IEEE Spectrum
https://spectrum.ieee.org/
With an improbable 850 kilowatts (1,140 horsepower), and a digitized launch from 0 to 100 kilometers per hour in a cerebellum-squeezing 2.5 seconds, the 2027 Porsche Cayenne Coupe Electric defies any traditional notion of an SUV. It’s the most powerful Porsche production car in history, thanks in part to DNA from Porsche’s championship Formula E racers. Automakers are notorious for using poetic license to draw tenuous connections between their racing cars and street cars. In this case, though, Porsche can draw legitimate links between their racers and their first-ever electric Cayenne . Exhibit “A” is the new Cayenne’s oil cooling of its rear electric motor, a direct transfer from its 99X racers . Another high-tech hand-me-down is its robust, 600-kilowatt braking-regeneration system, inherited from both the Formula E cars and Porsche’s Cayman GT4 ePerformance test car. “Formula E is our development lab for the electromobility of tomorrow,” said Michael Steiner , Porsche’s management board member for R&D, at the Cayenne’s unveiling last year. “The Cayenne Electric shows how quickly such a technology transfer takes place at Porsche.” I got to experience these innovations this past May during a scenery-blurring test drive south of Munich. I was especially impressed by the motor’s ability to halt this nearly 2,700-kilogram SUV from speeds as high as its electronically limited 261-km/h maximum (which, by the way, the Cayenne had no trouble reaching during my drive on unrestricted portions of the autobahn). That kind of power in stopping and energy recuperation traces to the race cars and their efficient thermal management. Push harder on the physical brake pedal, and the Porsche transitions from pure regenerative braking through its electric motor to activate its powerful friction brakes. For most drivers, those friction brakes won’t get much use, though: The Cayenne makes about 97 percent of its stops in typical driving using regenerative brakes alone, Porsche says. Those regenerative stops return up to 98 percent of available kinetic energy to the onboard battery, nearly matching the company’s Formula E racers for efficiency. How Oil Cooling Unlocks Massive Regenerative Stopping Power The oil-cooled motor is the key that unlocks such massive stopping power and energy recovery, without any risk of cooking internal components. Timo Henn, drive system manager for Porsche SUVs, pointed out the highlights in a cutaway model of the rear electric motor, developed at Porsche’s fabled Weissach R&D facility and built in Zuffenhausen. Heat is a big challenge in powerful motors, such as the ones used in high-performance electric vehicles (EVs). Excessive heat reduces efficiency by raising resistance. It also embrittles and degrades coil windings and can reduce the effectiveness of the motor’s lubricants. Improved cooling allows higher levels of current through the motor windings and therefore higher power. In a conventional EV motor, heat is typically dissipated from the stator by means of a water-glycol blend that flows through a cooling jacket outside the stator. But in this Cayenne machine, a non-electrically-conductive synthetic coolant oil flows directly along and through the motor’s live copper conductors, dissipating heat directly from its source. This oil, developed by Mobil , is called Therm Electric P , and it’s about five times thinner than conventional engine oil. That low viscosity allows it to flow freely through narrow gaps in the stator. The oil courses through its own cooling circuit, Henn says, and never needs replacement. As in some leading EVs, flat rectangular-cross-section winding conductors with hairpin curves replace round wiring, allowing densely packed copper to fill nearly 70 percent of the stator’s area, about 20 percent more than in traditional motors. Porsche does not cite the output of the rear motor alone, but rather a combined total of 850 kilowatts for the dual-motor configuration in all-wheel-drive mode. With 800 volts, the Cayenne’s PPE (Premium Platform Electric) architecture also enables exceptionally fast DC charging. Formula E racing is now known for its “Pit Boosts,” where a 30-second pit stop can boost the battery of the Porsche 99X racer by 10 percent in just 30 seconds. All told, the Porsche can charge its 108 kilowatt-hour battery from 10 to 80 percent in less than 16 minutes. It can add up to 328 kilometers (204 miles) of range in 10 minutes. Interrupting the test drive for a lunch stop on the Starnberger See, a glittering freshwater lake south of Munich, I sample another form of fast food, the Cayenne’s optional inductive charging system. I drive the Cayenne until it rests atop a magnetic ground pad that owners can wire into their electrical panels. An onboard screen guides me into docking position by aligning a pair of animated circles, one representing the moving Porsche, the other the pad. Within seconds, the Cayenne is slurping up 11 kilowatts of juice from the pad’s 85-kilohertz magnetic field, enough to go from 10 to 80 percent in less than eight hours—no grubby cord or bulky connector to deal with, and no forgetting to plug in. When IEEE Spectrum first covered the charger , my impression was that the €7,000 system (US $8,000) was quite pricey for consumers. Now I’m not so sure. The Turbo Coupe Cayenne I tested starts at US $170,350 and reaches $233,000 with options. Even a base-model Cayenne Coupe Electric starts at US $116,150. With Porsche buyers already spending those kind of sums, another €7,000 or $8,000 to cut the cord may feel like money well spent.
spectrum.ieee.org
1 day ago
A golf cart seems like an unlikely place to find a new model for industrial electrification, but a technical shift playing out on golf courses is pointing toward something larger. Swappable lithium battery packs are now replacing lead-acid units in legacy electric golf carts without any modification to the vehicles themselves: no controller swap, no wiring changes, no new charger required. That shift reflects a broader strategy taking hold across industries: engineering lithium battery packs that slot directly into older, smaller electric vehicles —golf carts, forklifts, scissor lifts, airport ground support equipment—without requiring those vehicles to be rebuilt or replaced. The appeal is straightforward: Lithium chemistry charges faster, lasts longer, and costs less to operate than lead-acid . But until recently, getting those benefits meant expensive modifications or buying new equipment entirely. Now battery makers are trying to change that by putting all the compatibility work inside the battery pack itself. Trojan Battery Company , headquartered in Santa Fe Springs, Calif., recently expanded compatibility for its OnePack lithium battery to include legacy versions of the E-Z-GO RXV golf cart. Dakota Lithium of Seattle, Relion Battery of Rock Hill, S.C., and Roypow Technology of Huizhou, China, are pursuing the same strategy across a wide range of vehicles and industrial equipment—all smaller than passenger EVs but collectively representing enormous markets. The global forklift market alone was valued at roughly US $82 billion in 2024, according to Grand View Research , a San Francisco-based market research firm. The aerial work platform market—which comprises machines such as scissor lifts , boom lifts , and vertical mast lifts , added another $11 billion that same year, according to Imarc Group , a strategy consulting firm headquartered in Noida, India. The size of those markets matters because the installed base of lead-acid-powered machines is not shrinking on its own. Warehouses, construction yards, and airports have no strong incentive to scrap equipment that still functions—which means the opportunity for drop-in lithium isn’t tied to new vehicle sales. It’s tied to the replacement cycle of batteries already in the field. That’s a different kind of market, and the companies pursuing it are betting it’s a larger one. Lithium Battery Retrofits for Golf Carts The obstacle to replacing lead-acid with lithium was electrical rather than mechanical. In a standard 48-volt lead-acid system, pack voltage starts near 50 volts when fully charged and declines into the low 40s as the battery empties. Older motor controllers use that descent as a signal, reading the voltage curve to regulate power delivery and estimate remaining charge. It was a design assumption so fundamental it was rarely documented, because no one expected it to change. Lithium packs break that assumption. They enter a discharge cycle at higher voltages—often in the mid-to-high 50s—and hold close to their nominal level before dropping sharply near the end of the cycle. That flat profile confuses controllers calibrated for a lead-acid curve, producing inaccurate state-of-charge readings, erratic performance, and abrupt shutdowns. Anyone who wanted lithium performance in a legacy machine faced a cascade of modifications: replacing the motor controller, installing a new charger, modifying battery mounts, reworking wiring. For large fleets, the economics rarely justified it. The Trojan Battery Company’s lithium batteries are designed to replace lead-acid battery packs for small vehicles with minimal fuss. Trojan Battery Company The solution was to push all of the complexity into the pack itself. As Darren Brittain , Trojan’s vice president of lithium commercial strategy for motive applications, says, the first step is “designing the lithium system around the electrical reality of the legacy platform, not asking the vehicle to adapt to the battery.” Trojan’s 48-volt, 105 ampere-hour OnePack is a 51.2-V lithium iron phosphate system with a working voltage range of 40.48 to 58.40 V, calibrated to stay within the tolerance of many legacy 48-V platforms. An onboard battery management system monitors cell voltages, balances charge, enforces thermal limits, and modulates output autonomously. Legacy machines read pack voltage, current demand, and charger behavior—nothing more. The battery management system handles everything else. “A good drop-in battery should not be the most aggressive lithium battery possible,” Brittain says. “It should be the best-performing lithium battery that the legacy platform can safely use.” The OnePack is rated at 180 amperes continuous discharge and 300 amps pulse—limits designed to keep the system inside an envelope that legacy vehicles can tolerate. That restraint is the point. The battery is engineered not to maximize what lithium can do in isolation, but to maximize what it can do inside a machine that was never designed for it. Lithium Forklift and Scissor Lift Retrofits The installed base of lead-acid-powered industrial machines is vastly larger than the world’s fleet of golf carts. Warehouses run electric forklifts kept in service for a decade or more. Construction and rental companies operate scissor lifts, boom lifts, and vertical mast lifts under the same constraints. Airports run baggage tugs, belt loaders, and cargo tractors on 24-V and 48-V systems unchanged for decades. Flux Power of Vista, Calif., has deployed more than 30,000 battery packs into forklifts, walkie pallet jacks, and airport ground support equipment that previously ran on lead-acid. Roypow and Trojan cover additional equipment categories as well: floor cleaning machines such as ride-on scrubbers and sweepers, and aerial work platforms such as scissor lifts and boom lifts. The operational advantages of making the switch are substantial. Lead-acid batteries take six to eight hours to charge and require an eight-hour cooldown before reuse, according to Relion. Lithium-ion batteries charge in one or two hours and can be topped off during breaks without affecting lifespan—a practice the industry calls “opportunity charging.” A single lithium pack can therefore power equipment through multiple shifts, whereas lead-acid systems require a dedicated battery for each shift, along with the space, infrastructure, and labor to manage the rotation. “‘Drop-in’ does not mean ‘universal with no validation.’... It means the system is engineered to minimize vehicle modifications while still requiring the right battery, charger, mounting hardware, firmware, and application match.” —Darren Brittain, Trojan Battery Company The long-term economics reinforce the case. Trojan’s OnePack delivers up to 4,000 charge cycles at standard operating temperature, replacing three or four lead-acid sets over a 10-year service life. For multishift operations, the transition typically delivers a return on investment within 36 months, according to LithiumLift , an Indianapolis-based forklift equipment and services company that compiled the figure from data drawn from hundreds of warehouse conversions. That timeline makes the decision relatively straightforward for high-utilization operations—the kind that run equipment around the clock and feel every hour of downtime directly in their margins. What makes that ROI figure significant is what it doesn’t require: new equipment purchases, retraining operators, or overhauling facility infrastructure. The machines stay. The batteries change. That simplicity is what separates the drop-in model from conventional electrification strategies, and why the companies pursuing it believe they can move faster than approaches that start from the vehicle up. Future Battery Chemistries and Legacy Gear The limits of the drop-in model are real, and the companies involved are careful to say so. “‘Drop-in’ does not mean ‘universal with no validation,’” says Brittain. “It means the system is engineered to minimize vehicle modifications while still requiring the right battery, charger, mounting hardware, firmware, and application match.” Compatibility claims must hold across different vintages, firmware versions, and usage profiles—a battery validated for one model year of a platform may behave differently in another. That validation work is ongoing, and it is not trivial. The challenge will only grow more complex as equipment ages through multiple technology generations. Vehicles that shipped with first-generation lithium systems are already reaching replacement cycles of their own. E-Z-GO RXV Elite models from around 2016 and 2017 are among them. The problem doesn’t disappear as equipment gets newer. It evolves. Each new battery generation may push voltage profiles, communication protocols, or packaging requirements further from what the previous generation assumed, narrowing the window in which engineering for legacy compatibility remains viable. Emerging chemistries add another layer of uncertainty. Sodium-ion and solid-state technologies are advancing, but chemistry is only part of the problem. The next battery must still match voltage windows, packaging, charging behavior, communication protocols, and safety requirements of the machines it replaces. A sodium-ion pack with superior energy density is useless in a legacy forklift if its discharge curve triggers fault conditions in the controller. The engineering burden doesn’t shrink as the technology improves—in some respects, it grows. That tension points toward the central bet the industry is making. When battery makers accept the burden of compatibility rather than passing it to the operator, the retrofit math changes, the upgrade decision simplifies, and a machine facing retirement gets a viable second act. At the scale of the global installed base of lead-acid industrial equipment, that’s not a marginal improvement, but a different theory of how electrification spreads. For decades, better battery technology meant buying new machines. That assumption is now the thing being replaced.
spectrum.ieee.org
1 week ago
The electric vehicle (EV) market in Europe is flourishing, to the point where, in late 2025, nearly 100 percent of new registrations in Norway were EVs . But large electric freight trucks, called electric heavy-goods vehicles (eHGVs), are still as rare as hen’s teeth. Researchers and innovators on the continent are seeking to change this picture, and fast. Chugging around the European Union are around 4.5 million HGVs , of which only around 14,500, or 0.32 percent, are electric. Figures in the United Kingdom are even more dire. Of the ~625,000 U.K. HGVs , a little over a thousand are eHGVs , making 0.16 percent. Despite diesel prices going through the roof since the start of the Iran war, eHGV sales are not trending upward. In late April, the U.K.’s Society of Motor Manufacturers and Traders (SMMT) reported a drop in eHGV registrations from 1.4 percent last year to 0.9 percent this year. Similarly, the E.U. market is stalling, climbing only from 4.2 percent of new registrations in 2025 to 4.4 percent so far this year . Why? The long-haul problem “On the last mile, people are very happy to switch to electric,” summarizes Alex Foote, of Heriot-Watt University, in Edinburgh, who leads the road part of the Transit project , a large-scale research program seeking to holistically decarbonize all U.K. transport—road, rail, maritime, and air—using digital twinning. “It’s long haul where there’s big range anxiety, there are big costs, and then we also have the ‘payload penalty.’” The payload penalty refers to how increasing an eHGV’s range calls for more batteries, whose weight cuts into the payload. One major improvement would be to speed up charging. A standard CCS2 (Combined Charging System Type 2) rapid charger delivering maximum 350-kilowatt power takes four hours to fully recharge an eHGV with a ~350-kilometer range, a completely impractical amount of time for most long-haul applications. The new Megawatt Charging System (MCS) —international standards for which were only fully ratified in early 2026—is designed to address this problem. The MCS can deliver over 1 megawatt of power, meaning it can charge a massive HGV battery in 30–45 minutes, perfect for a driver’s mandatory 45-minute break every 4.5 hours of driving. However, Foote sees practical flaws. “A lot of drivers say that they’re not on break if the vehicle is charging because they have to be there, monitoring it,” he says. “Also, it needs a very reliable and universal booking system, because drivers will need to know there’s a charger there with their name on it that’s not broken or in use.” On top of this, a truck stop or depot with 10 MCS chargers needs a 10 MW+ connection, equivalent to the needs of about 10,000 homes. Such a massive draw on the power grid could exceed local power constraints, and add massive cost. Thoughtful eHGV implementation Instead of installing huge MCS stations in every fleet operator’s depot and at every motorway service station, many researchers and innovators see a more realistic and practical way forward in better combining existing technologies. The U.K.-based battery innovator Zenobē says that requires thinking holistically. In 2017, a bus company complained that the cost of installing depot charging infrastructure would be more than the cost of the 10-bus fleet itself, and take three years. Zenobē came up with a more tailored solution that reduced this cost to half that of one vehicle, and completed the job in six weeks. The rear of an iONTRON electric ready-mix concrete truck (eMixer) fitted with a 350-kilowatt-hour battery, part of a trial project with the building-materials supplier Aggregate Industries. Zenobē “What we see going wrong in a lot of projects is you have ‘margin squirrelers’ across the supply chain, who are all looking for safety buffers because they’re not responsible for the total result,” says cofounder Steven Meersman. “In contrast, we take the attitude that this isn’t a vehicle problem, this isn’t a charging problem—it’s a problem that’s all about optimizing your whole operation.” Zenobē’s also addresses two other pain points for fleet operators. One is providing private financing options for fleet electrification projects when government grant funding is insufficient or unavailable, reducing initial outlays. The other is removing any risk surrounding battery-life degradation. Zenobē replaces batteries when their capacity is below a certain threshold, but then uses these old batteries for second-life applications. These batteries might find use as an alternative power source for eHGV charging during peak energy demand, a strategy called “peak shaving,” or they might be used to work alongside a diesel generator on construction sites. These second-life applications have real-world value, which Zenobē can then pass on to their customers. “This means that the [eHGV] customer only pays for what they use,” Meersman says. Software will drive eHGV adoption The Swedish freight-technology company Einride offers a somewhat different holistic solution for electrifying fleets. “Instead of thinking of the transition as a gradual electrification of an existing fleet, we took a step back and asked, ‘Where would full electrification make sense now?’” says electric-mobility general manager David Hallgren. “We wanted to start there and operate more or less entirely with an electric-only fleet from day one.” Einride’s fully autonomous, driverless, cab-less electric trucks have been operational on public roads since 2019, even completing the world’s first driverless international border crossing in 2025, between Sweden and Norway. But it’s the company’s Saga AI software that sets Einride apart. This software simultaneously weighs up all of the usual freight-operation factors as well as those specific to eHGVs, such as state of charge, sizes of loads, grid connections, topology, driving style, even the weather. It then learns from real-world data and applies it to future scenarios in order to continuously improve. Further improving Saga AI, Einride recently partnered with the U.S. quantum-computing company IonQ to help solve a nagging problem in freight logistics. Idle schedule gaps caused by shipment cancellations are difficult to fill optimally using classical optimization techniques. Combining classical techniques with a quantum approximate-optimization algorithm allowed the partners to achieve improvements of up to 12 percent in shipments delivered and a reduction of up to 6 percent in drive distance . “The number of factors you need to consider and the nonlinearity of how those intersect mean that it becomes impossible to manage an eHGV fleet at scale with any level of manual planning,” says Hallgren. “This is why we’re incorporating AI…and trying to look around the corner at new technologies that will allow us to do this even better.”
spectrum.ieee.org
1 week ago
Car thieves are getting better and better at exploiting security gaps in hands-free unlocking and keyless start systems. According to recent statistics from the U.K. , in fact, criminals hack cars’ keyless entry today more often than resorting to the old-fashioned, physical lock-bypass (a.k.a. slim jim ) method. And now a new generation of key fob and car-security-chip systems are being developed to stop tech-savvy car thieves cold. STMicroelectronics , based in Plan-les-Ouates, Switzerland, is one of several suppliers developing upgraded security chips for cars, smart locks, and other products. Last month the company released its ST64UWB line , designed in part to patch an exploit that thieves had previously used. The weakness in previous-generation systems concerns car locks that verify a key’s identity only, not the distance to that key as well. Even when automakers add ultra-wideband (UWB)—a secure, short-range wireless communication technology using high-frequency pulses to measure distance—the automakers often treat UWB tech as optional. When UWB distance measurements become unreliable, such as when a key is buried in a bag or in a pocket alongside other items, the security systems neglect distance data altogether. Enforcing Distance as a Security Signal Automakers and chipmakers are now trying to remove that loophole. Because while ultra-wideband chip technology is not new, what is changing is how reliably it works in real-world conditions—and whether vehicles can depend on UWB every time. Car thieves have developed exploits that rely on the loophole. One way to do so involves using two cheap transmitter-receiver radios that act as signal repeaters. A common scenario: A pair of crooks spies out a vehicle parked on a driveway and decides to take it. One stands right outside the front door, guessing that the car owner’s key ring is hanging on a wall near the entrance. His partner in crime stands near the vehicle. What makes this attack possible is that the key fob never goes silent—it regularly broadcasts a low-power radio signal, even when it’s sitting on a hook inside the house, because the car is always quietly asking whether a valid key is nearby. The thieves exploit that constant chatter between car and key fob: Their radios pick up the fob’s signal leaking through the wall and amplify the signal to a level that the car can pick up. The car receives what seems like a valid key right outside its door. And no button is ever pressed. No glass is broken or lock jimmied. The car accepts the relayed signal as proof of the key fob’s presence—although the fob is still just hanging on the hook inside the house. As the above example shows, signal strength can be manipulated with amplification or directional antennas . But the advantage of the new UWB system is that signal timing is much harder to spoof. Neal Patwari , a University of Utah professor of electrical and computer engineering who studies wireless networks and statistical signal processing , explains that attackers cannot make a signal transmitted from a key fob to a car’s security chip arrive sooner than the laws of physics allow. The thieves can only delay the signal—and delay makes the key fob appear farther away, not closer. That constraint is what allows automakers to treat proximity as a security check. “This moves security from ‘Is the key valid?’ to ‘Is the valid key physically close enough?’ ” says Rene Wutte , STMicroelectronics head of marketing for the company’s ranging and connectivity division. Instead of inferring distance from signal strength, ultra-wideband technology key fobs and car locks provide higher-security access to a vehicle based on signal travel time. STMicroelectronics What UWB Does—and Does Not Solve The new ultra-wideband chips make it difficult to fake proximity, but UWB does not eliminate all wireless vulnerabilities. Attackers can still jam the radio channel. As Patwari points out, interference can block a lock command when a driver presses the button, leaving the vehicle unlocked because it never receives the signal. That attack does not grant access. It prevents the security action of locking the door from completing. “But if my headlights don’t flash [indicating that the car has locked the doors], I’ll just push the button again,” says Patwari. “I won’t leave the car until I get that confirmation.” The scale of the problem is now visible in official data. As noted above, according to Crime Survey for England and Wales data published by the Office for National Statistics , 58 percent of vehicle thefts there involve keyless entry methods, including relay attacks. The Association of British Insurers corroborates that finding, linking a rising proportion of auto-theft claims to exploitation of keyless systems. Thatcham Research , which performs automotive risk analysis, and Tracker , which specializes in stolen vehicle recovery, also identify relay-style attacks as a leading method in modern vehicle theft. Mark Rose , managing director at Tracker, recently told the Insurance Times that, “As technology advances, organized criminal gangs develop increasingly sophisticated techniques to overcome existing vehicle security.” The remaining challenge is enforcement. If vehicles require distance verification every time, relay attacks become far harder to execute. If the systems still allow fallback—relaxing the physical proximity requirement that the improved UWB chips can enable—the vulnerability remains. Chipmakers like STMicroelectronics and their competitors are betting that improved reliability and increased security will make that choice easier.
spectrum.ieee.org
1 month ago
Why does a chocolatier build a railroad? For Milton S. Hershey, it was a logical response to a sugar shortage brought on by World War I. The Hershey Chocolate Co. was by then a chocolate-making powerhouse, having refined the automation and mass production of its products, including the eponymous Hershey’s Milk Chocolate Bar and the bite-size Hershey’s Kiss. To satisfy its many customers, the company needed a steady supply of sugar. Plus, it wanted a way to circumvent the American Sugar Refining Co., also known as the Sugar Trust, which had a virtual monopoly on sugar processing in the United States. Why Did Hershey Build an Electric Railroad in Cuba? Beginning in 1916, Hershey looked to Cuba to secure his sugar supply. According to historian Thomas R. Winpenny, the chocolate magnate had a “personal infatuation” with the lush, beautiful island. What’s more, U.S. business interests there were protected by a treaty known as the Platt Amendment , which made Cuba a satellite state of the United States. Like many industrialists of the day, Hershey believed in vertical integration, and the company’s Cuban operation eventually expanded to include five sugar plantations, five modern sugar mills, a refinery, several company towns, and an oil-fired power plant with three substations to run it all. A 1943 rail pass entitled the holder to travel on all ordinary passenger trains of the Hershey Electric Railway. Hershey Community Archives The company also built a railroad. To maximize the sugar yield, the cane needed to be ground promptly after being cut, and the rail system offered an efficient means of transporting the cane to the mills, and ensured that the mills operated around the clock during the harvest. By 1920, one of Hershey’s three main sites was processing 135,000 tonnes of cane, yielding 14.4 million kilograms of sugar. Initially, the Hershey Cuban Railway consisted of a single 56-kilometer-long standard gauge track on which ran seven steam locomotives that burned coal or oil. But due to the high cost of the imported fuel and the inefficiency of the locomotives, Hershey began electrifying the line in 1920. Although it was the first electrified train in Cuba, rail lines in Europe and the United States were already being electrified. In addition to powering the various Hershey entities, the generating station supplied Matanzas and the smaller towns with electricity. F.W. Peters of General Electric’s Railway and Traction Engineering Department published a detailed account of the system in the April 1920 General Electric Review . Hershey’s Company Towns The company town of Central Hershey became the headquarters for Hershey’s Cuba operations. (“Central” is the Cuban term for a mill and the surrounding settlement.) It sat on a plateau overlooking the port of Santa Cruz del Norte, about halfway between Havana and Matanzas in the heart of Cuba’s sugarcane region. Hershey imported the industrial utopian model he had established in Hershey, Penn., which was itself inspired by Richard and George Cadbury’s Bournville Village outside Birmingham, England. The chocolate magnate Milton S. Hershey had a “personal infatuation” with Cuba. Underwood Archives/Getty Images In Cuba as in Pennsylvania, Hershey’s factory complex was complemented by comfortable homes for his workers and their families, as well as swimming pools, baseball fields, and affordable medical clinics staffed with doctors, nurses, and dentists. Managers had access to a golf course and country club in Central Hershey. Schools provided free education for workers’ children. Milton Hershey himself had very little formal education, and so in 1909 he and his wife, Catherine, established the Hershey Industrial School in Hershey, Penn. There, white, male orphans received an education until they were 18 years old. Now known as the Milton Hershey School, the school has broadened its admission criteria considerably over the years. Hershey duplicated this concept in the Cuban company town of Central Rosario, founding the Hershey Agricultural School . The first students were children whose parents had died in a horrific 1923 train accident on the Hershey Electric Railway. The high-speed, head-on collision between two trains killed 25 people and injured 50 more. Milton Hershey was a generous philanthropist, and by most accounts he truly cared for his employees and their welfare, and yet his early 20th-century paternalism was not without fault. He was a fierce opponent of union activity, and any hard-won pay increases for workers often came at the expense of profit-sharing benefits. Like other U.S. businessmen in Cuba, Hershey employed migrant seasonal labor from neighboring Caribbean islands, undercutting the wages of local workers. Historians are still wrangling with how to capture the long-lasting effects of U.S. economic imperialism on Cuba. Can the Hershey Electric Railway Be Revived? Hershey continued to acquire new sugar plantations in Cuba throughout the 1920s, eventually owning about 24,300 hectares and leasing another 12,000 hectares. In 1946, a year after Milton Hershey’s death and amid growing political uncertainty on the island, the company sold its Cuban interests to the Cuban Atlantic Sugar Co. In addition to Hershey’s sugar operations, the sale included a peanut oil plant, four electric plants, and 404 km of railroad track plus locomotives and train cars. Service on the Hershey Electric Railway in Cuba continued into at least the 2010s but became increasingly sporadic, with aging equipment like this car at the Central Hershey station. Hershey Community Archives The Central Hershey sugar refinery continued to operate even after the Cuban Revolution but eventually closed in 2002. Passenger service, meanwhile, continued on the Hershey Electric Railway, albeit sporadically: By 2012, there were only two trips a day between Havana and Matanzas. This video, from 2013, gives a good sense of the route: A colleague of mine who studies Cuban history told me that in his travels to the country over almost 30 years, he has never been able to ride the Hershey electric train. It was always out of service or had restricted service due to the island’s chronic electricity shortages , which have only gotten worse in recent years. I’ve been trying to find out if any part of the line is still operating. If you happen to know, please add a comment below. Cuba’s frequent power outages make it difficult to operate the Hershey Electric Railway. In this 2009 photo, passengers await the restoration of electricity so they can continue their journey. Adalberto Roque/AFP/Getty Images A 2024 analysis of the economic potential and challenges of reactivating Cuba’s Hershey Electric Railway noted that an electric railway could be a hedge against climate change and geopolitical factors. But it also acknowledged that frequent power outages and damaged infrastructure argue against reactivating the electrified railway, and it favored the diesel engines used on most of Cuba’s rail network. Cuba has been mostly off-limits to U.S. tourists for my entire life, but it was one of my grandmother’s favorite vacation spots. I would love to imagine a future where political ties are restored, the power grid is stabilized, and the Hershey Electric Railway is reopened to the Cuban public and to curious visitors like me. Part of a continuing series looking at historical artifacts that embrace the boundless potential of technology. An abridged version of this article appears in the May 2026 print issue as “This Chocolate Empire Ran on Electric Rails.” References In April 1920, F.W. Peters of General Electric’s Railway and Traction Engineering Department wrote a detailed account called “ Electrification of the Hershey Cuban Railway ” in the General Electric Review, which was later abstracted in Scientific American Monthly to reach a broader audience . Thomas R. Winpenny’s article “ Milton S. Hershey Ventures into Cuban Sugar ” in Pennsylvania History: A Journal of Mid-Atlantic Studies, Fall 1995, provided background to the business side of Hershey’s Cuba enterprise. Florian Wondratschek’s 2024 article “ Between Investment Risk and Economic Benefit: Potential Analysis for the Reactivation of the Hershey Railway in Cuba ” in Transactions on Transport Sciences brought the story up to the present. And if you’re interested in a visual take on the Hershey operation on Cuba, check out the documentary Milton Hershey’s Cuba by Ric Morris, a professor of Spanish and linguistics at Middle Tennessee State University.
spectrum.ieee.org
1 month ago
This article is part of our exclusive IEEE Journal Watch series in partnership with IEEE Xplore. Rail networks are vast, which makes it difficult to conduct comprehensive, continuous safety monitoring. Researchers in China have suggested analyzing the vibrations of existing fiber cables buried underground alongside railway tracks to detect problems. In a study published 5 March in the Journal of Optical Communications and Networking , the research group demonstrated through experiments how the technique can successfully identify a number of issues associated with train safety, including faulty train wheels and broken sound barriers alongside the railway tracks. Sasha Dong is a junior chair professor in Southeast University’s School of Transportation, in Nanjing, China. She notes that traditional approaches for monitoring railways—such as video surveillance, radar, and ultrasonic sensing—can be effective, but they are often limited to monitoring railways at single points along entire systems. “As a result, they are not well suited for continuous coverage along an entire railway line and are also more vulnerable to weather conditions, environmental factors, and power supply constraints,” she says. Instead, Dong, Yixin Zhang at Nanjiang University, and their colleagues used a technique called distributed acoustic sensing (DAS) to analyze the vibrations of underground optic fiber cable alongside railway tracks to detect safety issues. Specifically, pulsed light is sent along the cable, and the propagation of scattered light is used to detect and quantify vibrations along the cable. The researchers developed AI models to filter out the noise from those signals and to identify the particular vibrations associated with various kinds of unsafe conditions, such as damaged or defective wheels. Dong notes that railways already have extensive optical fiber networks for communication buried underground alongside them, meaning that the cables can be harnessed as a sensing medium with no extra power supply or need for another expensive network to be constructed. Instead, monitoring stations could be installed at intervals along the railway track, with extension cables connecting a DAS system to the main cable. Machine Learning for Railway Safety To develop their DAS system, the researchers set about collecting data on different railway safety issues and training machine learning algorithms to identify specific vibrations associated with each one. For example, they trained a model to detect the trajectory of trains using DAS data. This involved more than 13,000 samples of trains moving along tracks, where their direction was confirmed using data. This model achieved an accuracy of 98.75 percent. In another endeavor, the researchers took samples of a train with wheel-pair faults—where there is damage or a defect on the railway wheels or their connecting axle—moving along a 60-kilometer stretch of railway track in Kunming, Yunnan, China. The researchers were able to clearly detect when there was an issue: The vibration frequencies of normal wheels were mainly concentrated below 60 hertz, while the frequency of faulty wheels could get as high as 100 Hz. DAS may also be useful for detecting problems with sound barriers, which are the paneled walls on either side of the railway track that reduce the sound of trains as they pass surrounding neighborhoods. The researchers removed the rubber paneling from sound barriers to simulate faulty barriers and repeatedly struck the barrier with a rubber hammer, using the resulting sound data to train another model. This model could accurately detect faulty sound barriers with 99.6 percent accuracy. The team also explored how well machine learning algorithms could detect abnormal events along the railways, such as humans climbing over trackside fences, rocks falling on the track, illegal construction activity such as excavator operations, or other environmental disturbances. These types of events were a bit more difficult to distinguish at first, but by feeding a lot of data into the model, the researchers were able to boost the model’s accuracy for these types of events to 97.03 percent. These results suggest that DAS has the potential to be an effective tool for monitoring railway systems. “What we have found most surprising is that a single, existing fiber deployed along the railway, with appropriate modeling and algorithm design, can support so many different monitoring tasks at the same time,” says Dong. “This kind of multipurpose use of one fiber system has strong engineering value.” Dong acknowledges that these experiments were done in controlled environments and emphasizes the need to capture more vibration data under real high-speed train operating conditions. Nevertheless, she says, “the results of this study suggest that this [approach] is feasible and has strong potential for practical application.” This article appears in the June 2026 print issue as “Optical Fibers Sense Risks on the Rails.”
spectrum.ieee.org
1 month ago
Drivers in 46 U.S. states were left temporarily stranded in March when breathalyzer devices installed on their cars went inactive. These ignition-interlock devices require a driver, typically one who has been convicted of an alcohol-related driving offense, to breathe into a tube and will prevent the car from starting if the system detects alcohol beyond a state-mandated threshold that’s well below the standard legal blood alcohol limit. A cyberattack on backend systems at Intoxalock , an ignition-interlock provider , prevented devices from verifying compliance. The episode highlights how failures in cloud services can reverberate through the physical world of mobility. Beginning around 14 March, users reported devices rejecting startup attempts or blocking required logins. Affected users took to social media to complain of missed work shifts, canceled appointments, and cars stuck in driveways—even when drivers successfully passed breath tests. Intoxalock, one of the largest ignition-interlock providers in the United States, later confirmed a cybersecurity incident affecting portions of its backend infrastructure. The company said vehicle-safety systems were not compromised, and breath-alcohol measurements remained accurate. The disruption instead stemmed from service outages that prevented service technicians from accessing the in-car systems’ software or routing maintenance logs certifying that the breathalyzers’ sensors were properly tuned. Unless the servers periodically receive up-to-date information on a car, they will not send the signal giving the ignition interlock permission to start the car’s engine. Compliance Systems Meet Connected Infrastructure Ignition-interlock devices periodically connect to backend servers to verify account status, upload logs of breath-test events, and confirm calibration schedules required for legal certification. “Local fallback is critical” –Sam Abuelsamid, Telemetry mobility research When backend communication failed during the cyberattack, some of the devices entered operating states designed to enforce court-mandated compliance rules. In these states, system software would not authorize ignition. Affected drivers reported their cars were out of commission from a couple of hours to several days. On 18 March, Intoxalock issued a notice to users advising them of a temporary workaround and a promise to compensate them for financial losses incurred because of the incident. The incident highlights a growing problem in automotive cybersecurity : Systems can fail and render perfectly functional vehicles useless. Sam Abuelsamid , vice president of market research at Telemetry , a Detroit-area transportation and mobility research and advisory firm, says the outage reflects a broader industry shift toward connected, software-dependent vehicles. “We’re absolutely getting into a situation where more and more vehicle services rely on connectivity or remote services in order to function,” he says. The trend goes back at least to 2019, when Tesla introduced a system allowing drivers to unlock and start their cars using a smartphone rather than a physical key or key fob. In areas where cellular coverage is weak or nonexistent, Abuelsamid notes, drivers who lacked a backup key could find themselves locked out of their cars. The downside of connectivity dependence in mobility is also apparent, he says, when robotaxi fleets operated by companies such as Waymo can’t link up with the servers they depend on for dispatch and supervision. And in China, roughly 100 autonomous vehicles operated by Baidu’s Apollo Go robotaxi service suddenly stalled in the middle of traffic in Wuhan on 31 March. Some passengers there were stranded for two hours. (Baidu later referred to the incident as a “system failure,” offering no details.) Connectivity and the Recurring-Revenue Model Automakers are increasingly outfitting cars with software features and connected functionality that owners can only access through regular payments. These subscription fees let the car companies break out of the traditional one-time sales model, but they also deepen reliance on backend infrastructure and increase the risk of outages due to communications breakdowns. To be sure, these software-dependent features deliver significant benefits, including over-the-air updates, remote diagnostics, and faster deployment of security patches. But they come at the price of greater system complexity and additional operational dependencies. Some automakers already incorporate backup access methods to reduce dependence on continuous connectivity. Ford , for example, complements its phone-as-a-key feature with a locally verified numeric PIN that ensures operability even when cloud services or mobile connections fail. Basically, the driver door has a keypad that can be used to unlock the car. The vehicle can then be started by entering the PIN on the dashboard touchscreen. “That kind of local fallback is critical,” Abuelsamid says. Similar local-processing safeguards, he notes, could reduce dependence on uninterrupted network access for compliance technologies. A Preview of Connected-Vehicle Risk The Intoxalock incident illustrates how outages affecting back-end systems serving vehicles dependent on apps, payment platforms, and cloud identity systems can propagate across large user populations simultaneously—more like a network failure than a traditional recall. Intoxalock said services were progressively restored as backend systems returned online, though the company has not released any further technical details about the attack or its resolution. But one thing is clear: As vehicles become increasingly software-dependent, resilience will depend as much on backend architecture as on vehicle hardware.
spectrum.ieee.org
1 month ago
When JoeBen Bevirt founded the company now known as Joby Aviation in 2009, electric aircraft were on few people’s minds. Tesla’s first electric vehicle, the Roadster, had arrived on the market just one year earlier, and it was not yet evident that EVs would transform the automotive industry, let alone aviation. But Bevirt was convinced that he could revolutionize urban transportation with electric air taxis, also known as electric vertical take-off and landing aircraft (eVTOLs for short). In a departure from many conventional aircraft manufacturers, Bevirt embraced a rapid, iterative design process that Joby refers to as “design, build, and test.” According to Jon Wagner , who spent five years as senior director of battery engineering for Tesla before joining Joby in 2017, “the original concept was that the more times you go through the process, each time you can identify improvements or problems that need to be addressed. And if you have a very good system for going through that iterative process quickly and at a low cost, then you can take risk, because if it fails, you just go again.” With no existing supply chain for electric aircraft, Joby applied this iterative design strategy to almost every component of its eVTOL, including its all-important electric power train. Joby cycled through several generations of a geared electric motor before Wagner joined the company, leading development of a direct-drive motor with superior reliability, performance, and noise characteristics. With Joby’s aircraft now undergoing certification with the U.S. Federal Aviation Administration , IEEE Spectrum caught up with Wagner to learn more about the company’s power-train technology and where he expects it to go in the future. Our interview with him has been edited for concision and clarity. What can you tell us about the design of Joby’s electric motor? Wagner: It’s a direct-drive motor running the propeller. That direct-drive motor has a fairly large diameter in order to get the high torque density that we want. Essentially, the challenge there is, we want to spin the propellers relatively slowly compared to most aviation propellers, to reduce the sound. The trade-off there is when you spin them slowly, you need a lot of torque, and so we have a fairly high-torque direct-drive motor. That’s arranged as a single magnet ring—the spinning part is the magnet ring; it’s well integrated onto the propeller system. The stationary portion is typically called the stator. For us it’s copper coils and magnetic steel that focuses the electromagnetic forces. That is made up of essentially two motors, so the stator is divided into two separate sets of coils, each driven by separate inverters, each sourced by separate batteries for redundancy. At a high level, that’s the layout. The details are how you separate those coils electrically, how you separate the inverters both physically and electrically, and then, really, what stitches all of this together is a thermal system. And the thermal system is very key to achieving low weight. Essentially, when we talk about the innovation we bring, it’s typically not invention, it’s integration. We typically pick solutions that the world knows about—there’s very little here that’s Ph.D. thesis work. It’s integration work. How important is manufacturability in the design of your motors, and does the fact that Joby is building its motors in house change your calculus there? Wagner: It really does. Table stakes for a good design is manufacturability, so it’s fundamental, and it was architected into the entire concept. Where the vertical integration that Joby has developed creates an advantage for us is actually exactly in manufacturability and how it pertains to performance. In a mature industry like the automotive industry, you typically see a very diverse supply chain and a very advanced kind of outsourced supply-chain methodology . The most efficient way to run these very mature companies is to break your system up into pieces that can be outsourced to suppliers who are going to do a really good job of each piece. And that actually works really well once the supply chain is mature, and they know how to make all those parts, and also once the industry knows what they need from each part. The downside is that when you break a problem up into three pieces, you now have interface boundaries between each of these pieces, and those interface boundaries always create inefficiencies, and that typically manifests as either complexity of the design or mass or potentially introducing reliability [concerns]. Here’s where the vertical integration comes in. We were able to design highly integrated solutions without taking the manufacturing penalty that would come from finding a supply chain where they would make each part to integrate to each other. And so by having that vertical integration in our team, we essentially address the manufacturing challenges, and we get the mass and performance benefits of a highly integrated solution. Where Superconducting Motors Could Matter Aviation motors are a hot topic in machine design, and there’s been considerable interest in the use of superconductive or carbon nanotube coils, for example. Are you exploring any of these? Wagner: Superconducting is a very interesting vector, and there is a lot of interesting work going on with superconducting materials. Essentially the win here is reducing the losses; losses are heat generation and energy that doesn’t turn into useful work for the airplane. And the first thing I’ll say is that motors without superconducting materials actually are quite efficient. We’re talking low to mid 90 percent efficient already. With these kinds of efficiencies, the win is somewhat small, and the effort is very high. And so we have looked at superconducting motors , but we’re not really working actively on this right now. Where this really gets interesting is often in a larger size, so bigger scale. We live in the hundreds of kilowatts scale. When you get to the multiple megawatts kind of scale, 1 to 10 megawatts, for motors in that size, there could be some big advantages. Five percent of 10 megawatts ends up being a lot, and so you could potentially get big gains there. There are challenges. The challenges with superconducting motors are that you need to get them very cold in order to maintain the superconducting features. And then if you want to take advantage of that superconducting performance, you essentially have a motor that if you lose the ability to keep it cold, it has almost no productive use. So in other words, it was never designed to work when it isn’t superconducting, and therefore your cryogenic cooling system, if there’s any failure in that system, you essentially cannot produce any power anymore, and so you end up being very reliant on the cryogenic cooling system. Potentially even the motor could fail very rapidly after a failure of the cooling system. These are solvable problems, but they bring challenges, and that’s part of the reason why I think we’ll see this first with the largest motors. How do you expect Joby’s power-train solutions to evolve in the future, balancing the availability of new technologies and materials against the difficulty of certifying new designs? Wagner: We essentially set out more than 10 years ago to build an airplane that demonstrates a shift out of fossil fuels, and the best way to do it at that time—and it turns out, still now—is with batteries. And so we’re now at the late stages of development into the certification, in the stable-design state. We essentially know it’s going to work, right? And that’s very exciting, because I think what we’ll see from here as we branch forward is continual improvement in batteries. We know that battery energy density is a huge problem. It’s very heavy for the amount of energy stored. It’s much worse than for fossil fuels, so right now, and probably well into the future, battery-based energy storage for aviation has somewhat limited market potential. We don’t really see batteries being able to handle long-haul. Those kinds of trips that most of us utilize, I don’t think we’ll see battery energy storage for those anytime in the near future. It’s no secret that Joby has a very active effort going with hydrogen as the primary energy source. We’ve been doing that for many years. We look at taking hydrogen and using fuel cells to convert that hydrogen to electricity and then essentially using the same electric propulsion system downstream. We think that’s a very exciting direction, both for regional transport and long-range transport, but especially for long-range transport, really the best solution. There are challenges that we’re actively working to solve. Number one, how do you store the hydrogen? Number two, how do you refuel and what does the ground infrastructure look like to get that source of fuel from the source of hydrogen onto an airplane? And then, number three, how do you convert that hydrogen to electricity? Based on what I see, the future of aviation absolutely comes down to storing energy, and hydrogen is fundamentally, at a molecular level, about three times better than fossil fuels at storing energy per mass. And that’s a really big deal. The opportunity is so big on hydrogen; when you have three times better on something like the storage density, it means that it’s worth putting a lot of research into how to make this work, because that’s going to pay off at the very end. And so we’re firmly in it for that. I’m very excited about that future, which yet may be decades off but has to start somewhere.
spectrum.ieee.org
2 months ago
Large language models have already transformed software engineering, for better or worse. Now, so-called large physics models are also starting to transform design engineering. These tools are beginning to replace—or at least amend—the role of full-fledged physics simulation in the automotive and aerospace industries, semiconductor engineering, and more. Before the advent of computer simulation, a car manufacturer, for example, would create prototypes to test their designs, says Thomas von Tschammer , managing director at physics-based AI company Neural Concept . “For the past 40 years, we reduced a lot of the need for prototypes by using numerical simulations for aerodynamics, for crash testing, and so on.” Now, von Tschammer explains, AI is drastically reducing the need for simulation, the same way simulation reduced the need for physical prototypes. Growing adoptions of this type of AI was a topic of interest at Nvidia GTC in March. Chris Johnston , senior technical specialist at Jaguar Land Rover, presented how his company is using Neural Concept’s technology. PhysicsX , another physics-based AI company, announced a collaboration with Nvidia to advance open standards for such models, also at GTC. The AI design engineering workflow Over the past six months, General Motors (GM) has introduced large physics models into their car design process to speed up the workflow. Previously, a creative design engineer would develop a 3D model of a new car concept. This model would be sent to aerodynamics specialists, who would run physics simulations to determine the coefficient of drag of the proposed car—an important metric for energy efficiency of the vehicle. This simulation phase would take about two weeks, and the aerodynamics engineer would then report the drag coefficient back to the creative designer, possibly with suggested modifications. Now, GM has trained an in-house large physics model on those simulation results. The AI takes in a 3D car model and outputs a coefficient of drag in a matter of minutes. “We have experts in the aerodynamics and the creative studio now who can sit together and iterate instantly to make decisions [about] our future products,” says Rene Strauss , director of virtual integration engineering at GM. For GM and other companies, running inference on an AI model trained on physics simulations, instead of running the simulation itself, can bring immense time savings. “Depending on the kinds of physics [being simulated], or the resolution, it can be anywhere between 10,000 to close to a million times faster,” says Jacomo Corbo , CEO and co-founder of PhysicsX. How accurate are large physics models? But what about accuracy? For GM’s purposes, Strauss says accuracy is not a huge concern at the design stage because finer details are ironed out later in the process. “When it really starts to matter is when we’re getting close to launching a vehicle, and the coefficient of drag is going to be used for our energy calculation, which eventually goes to the certification of our miles per gallon on the sticker.” At that stage, Strauss says, a physical model of the car will be put into a wind tunnel for an exact number. PhysicsX’s Corbo argues that, with the right data, the AI model accuracy can supersede the accuracy of the simulation it’s trained on. The trick is to incorporate experimental measurements to fine-tune the model. If a physics simulation doesn’t agree exactly with experimental data, it is often difficult to figure out why and tweak the model until they agree. With AI, incorporating a few experimental examples into the training process is a lot more straightforward, and it’s not necessary to understand where exactly the model went wrong. All in all, by drastically bringing down the time it takes to model the physics, large physics models enable engineers to explore a much greater range of possibilities before a final design is reached. Training large physics models There is no one-size-fits-all approach to training large physics models. Depending on the types of data available, and the physics in question, the models may use the transformer architecture that underlies LLMs, a generalized version of convolutional neural networks known as geometric deep learning , or an architecture that can solve partial differential equations called neural operators . Currently, most companies are training their own models on their simulation data, catering to specific use cases. In GM’s aerodynamics implementation, there are different AI models for different types of cars: think SUVs versus sedans. But PhysicsX’s Corbo says his team is working on building more “foundational” physics models that can be applied across different scenarios. Both LLMs and robotics have benefited from scaling laws, which describe how a system improves as the models increase in size or get trained on more data. In AI, models tend to improve quickly, in a nonlinear way. Along the way, the models also become more generalizable—extending them to new settings takes less and less fine-tuning to reach the same accuracy. Corbo says his team is now starting to see the same types of scaling laws for large physics models. “What we’re seeing here is maybe a little bit unsurprising,” Corbo says, “but it’s also pretty incredible. And it’s given us the confidence to make these models bigger, because they perform a whole lot better, and they cover broader domains, and they have these really amazing emergent properties.” Developing open standards for the data formats used in training, as well as the model architectures, should help develop these more powerful foundational models. That’s the goal of PhysicsX’s collaboration with Nvidia, and of Nvidia’s physicsNeMo open source platform. “The thing that we’re collaborating on is being able to compose architectures from building blocks,” Corbo says, making it easy for those in both academia and industry to reuse and build upon existing models. A type of AI called a large physics model is used by an engineer to quickly generate heat flow in a 3D data center server design. Neural Concept The long-term role of simulations and engineers While some are working on developing more powerful models, others are pushing to implement what’s already available into existing workflows, which is no easy task. “With any innovation, it’s not a straight line. There’s some steps forward and then some steps back and improvements that we find along the way. But that’s part of the joy of the innovation process and using new tools like this,” GM’s Strauss says. This technology is still in the early stages, and it’s unclear what the final role of AI tools will be in the engineering workflow. For one, opinions vary on whether AI will replace simulations completely, or just reduce their use. “We will never fully replace simulations,” Neural Concept’s von Tschammer says. “But the idea is to make a much smarter usage of simulation at the most major phase of developments, and you use AI to speed up the early design stages, where you need to explore a very wide set of options.” PhysicsX’s Corbo begs to differ. “The whole idea is to take numerical simulation … out of the workflow,” he says, “and to move that to inference.” Whatever the role of simulation will be, everyone in the field is adamant that human design engineers will continue to be in the driver’s seat, enabled by these newfangled tools to do their best work. (After all, when has AI ever threatened to replace human labor?) “What we’re seeing is that actually, these tools are empowering the engineers to be much more efficient,” von Tschammer says. “Before, these engineers would spend a lot of time on low-added-value tasks, whereas now these manual tasks from the past can be automated using these AI models, and the engineers can focus on taking the design decisions at the end of the day. We still need engineers more than ever.”
spectrum.ieee.org
2 months ago
Check out the interior of the self-driving ca r in Spielberg’s Minority Report that whisks Tom Cruise’s character toward jail: There are only two seats. Perhaps taking a page from that sleekly designed sci-fi, Lucid Motors revealed the Lunar , a hyperefficient robotaxi concept, at its recent Investor Day in New York City. With its two side-by-side seats, compact size, and a cabin freed from a steering wheel, pedals, and garrulous cabbie, the Lunar defies more than a century of taxi tradition. Lucid, which has partnered with Uber to deploy up to 20,000 of its seven-passenger Gravity SUVs as robotaxis, says that as many as 90 percent of taxi trips involve one or two passengers. Since passengers almost never sit up front in a human-driven taxi, having two rows of seats in this energy-saving model makes little sense, says Zach Walker, Lucid’s chief of advanced product creation. “People already view the front seat of a taxi as a no-go land,” he declares . The Lunar is a scaled-down version of Lucid’s forthcoming midsize Cosmos and Earth SUV’s. Walker explains that for the project his team was freed for a “technical moonshot” that could make this car among the world’s most energy-efficient production EVs. That kind of efficiency could be critical for a fledgling robotaxi business that seeks to squeeze every kilowatt and penny from cars that could might be cruising up to 20 hours a day, seven days a week. The Cosmos, a Tesla Model Y competitor, is no slouch, at up to 7.24 kilometers (4.5 miles) of driving range for every kilowatt-hour of battery energy, thanks to its new Atlas power train and a class-best 0.22 coefficient of drag . The Lunar advances the company’s goal of “radical efficiency” by further shrinking its battery size, to about 55 kilowatt-hours, down from 69 kWh in the Cosmos. Walker says the Lunar could deliver up to 9.7 kilometers (6 miles) of driving range for every kilowatt-hour of battery—nearly double the efficiency of a typical four-seat electric SUV. A quick calculation suggests that would be enough to travel more than 500 kilometers (310 miles) on a charge, despite the Lunar’s relatively pint-size battery. Downsizing Can Be a Virtuous Circle Downsizing batteries is a design tactic expounded by Lucid founder and former CEO Peter Rawlinson . He believed it sets off a virtuous circle or “convergent series” of efficiency gains, allowing less nonactive battery-pack material, supporting structures, and downsized brakes and suspension components. In other words, each weight reduction means that slightly less battery can deliver the same driving range. Up to a point, anyway. Sam Abuelsamid, an engineer and vice-president of market research for Telemetry , agrees the weight of a power train or battery can lead to a virtuous—or vicious—circle in engineering. “A Hummer EV is the worst example on the electric side, carrying almost 3,000 pounds of battery, but also all the structure (and associated components) to support it,” he notes. Taxis have traditionally been big, lumbering, and fuel-thirsty. Consider the iconic yellow cabs that Checker Motors built in Michigan from 1922 to 1982, or London’s tall-roofed hackney cabs, originally designed to provide head room for men’s top hats and bowlers. But today, Abuelsamid says, two-passenger robotaxis make obvious sense for urban areas where they are most likely to proliferate. “They have a smaller footprint, use less energy, and reduce congestion in cities,” Abuelsamid says. “You just wouldn’t want them for your entire fleet.” Efficiency gains can pay special dividends in robotaxis, which some industry leaders envision logging up to 100,000 miles a year. For every 1 kWh reduction in battery size, Walker calculates, that robotaxi workhorse would save up to $1,000 a year in operating costs. Lucid says the Lunar could reduce operating costs by 40 percent versus larger robotaxis retrofitted from passenger cars, such as Waymo’s Jaguar iPace models. Regarding charging, the larger Cosmos can already add 200 miles of range in 14 minutes on a DC fast charger. With its superior per-kilometer efficiency, the Lunar could likely add 200 miles in closer to 10 minutes, reducing service downtime that’s another critical calculation for taxi operators. At Investor Day in New York City, Lucid’s interim CEO March Winterhoff and Uber President Andrew Macdonald sat inside a Lunar concept car, which was shown with no doors—the better to flaunt its 36-inch display screen and spacious cabin. The Lunar integrates a large array of sensors to create a bird’s-eye view of its environment, including lidar, cameras, and radar. It’s powered by Nvidia’s new Drive Thor system-on-a-chip, designed to support Level-4 or Level-5 autonomy with 1,000 teraflops of compute performance for critical inference processing . Dispensing With the Giggle Factor Where Lucid’s Air and Gravity models are known for blistering acceleration and sporty handling, a utilitarian robotaxi has no need for “the giggle factor,” as Walker dubs it. That creates more opportunities for savings, and passenger comfort. A chassis can be optimized for a comfy ride and low NVH (noise, vibration, and harshness). Meanwhile, driver pedals, a steering wheel and complex linkages, and electrified assists are all eliminated. Dynamic steering, beefed-up body control or massive wheels and tires to boost cornering? No need. After all, there’s no human driver to experience those sensations. And a taxi passenger’s worst nightmare is a driver who thinks he’s Max Verstappen . Of course, robotaxis bring their own set of tech challenges. According to Walker, a current robotaxi might use up to 24 kWh of energy over 20 hours to sense its environment and operate safely. Most of that goes to processors and onboard sensors, with lidar an especial energy hog. Though the Lunar remains a concept for now, it’s no sci-fi fantasy. The Lunar was designed to use the same components front and rear as other midsize Lucids, differing only in its downsized battery and center passenger section. No complex, costly reengineering is required, and the Lunar could share a production line with those showroom SUVs. For all those reasons, Walker says the Lunar is fundamentally sound and ready to scale. All Lucid needs are customers. “We still have our day jobs, but this was like our midnight project that we were all obsessed with making,” Walker says. “We think the [robotaxi] industry is primed for a really cool takeoff.”
spectrum.ieee.org
2 months ago
As a kid, I loved the 1980s aquatic adventure show Danger Bay . True to the TV show’s name, danger was always lurking at the Vancouver Aquarium, where the show was set. In one memorable episode, young Jonah and a friend get trapped in a sabotaged mini-submarine, and Jonah’s dad, a marine-mammal veterinarian, comes to the rescue in a bubble-shaped underwater vehicle. Good stuff! Only recently—as in when I started working on this column—did I learn that the rescue vehicle was not a stage prop but rather a real-world research submersible named Deep Rover . What Was Deep Rover and What Did It Do? Built in 1984 and launched the following year, Deep Rover was a departure from standard underwater vehicles, which typically required divers to lie in a prone position and look through tiny portholes while tethered to a support ship. Deep Rover was designed to satisfy human curiosity about the underwater world. As the rover moved freely through the water down to depths of 1,000 meters, the operator sat up in relative comfort in the cab, inside a clear 13-centimeter-thick acrylic bubble with panoramic views—an inverted fishbowl, with the human immersed in breathable air while the sea creatures looked in. Used for scientific research and deepwater exploration, it set a number of dive records along the way. Submarine designer Graham Hawkes [left] and marine biologist Sylvia Earle [right] came up with the idea for Deep Rover . Alain Le Garsmeur/Alamy The team behind Deep Rover included U.S. marine biologist Sylvia Earle and British marine engineer and submarine designer Graham Hawkes . Earle and Hawkes’s collaboration had begun in May 1980, when Earle complained to Hawkes about the “stupid” arms on Jim, an atmospheric diving suit ; she didn’t realize she was complaining to one of Jim’s designers. Hawkes explained the difficulty of designing flexible joints that could withstand dueling pressures of 101 kilopascals on the inside—that is, the normal atmospheric pressure at sea level—and up to about 4,100 kPa on the outside. But he listened carefully to Earle’s wish list for a useful manipulator. Several months later, he came back with a design for a superbly dexterous arm that could hold a pencil and write normal-size letters. Earle and Hawkes next turned to designing a one-person bubble sub, which they considered so practical that it would be an easy sell. But after failing to attract funding, they decided to build it themselves. In the summer of 1981, they pooled their resources and cofounded Deep Ocean Technology, setting up shop in Earle’s garage in Oakland, Calif. Phil Nuytten, a Canadian designer of submersibles and dive systems, engineered Deep Rover . Stuart Westmorland/RGB Ventures/Alamy They still found that customers weren’t interested in their crewed submersible, though, so they turned to unmanned systems. Their first contract was for a remotely operated vehicle (ROV) for use in oil-rig inspection, maintenance, and repair. Other customers followed, and they ended up building 10 of these ROVs. In 1983, they returned to their original idea and contracted with the Canadian inventor and entrepreneur Phil Nuytten to engineer Deep Rover . Nuytten didn’t have to be convinced of the value of the submersible. He had grown up on the water and shared their dream. As a teenager, he opened Vancouver’s first dive shop. He then worked as a commercial diver. He founded the ocean- and research-tech companies Can-Dive Services (in 1965) and Nuytco Research (in 1982), and he developed advanced submersibles as well as diving systems. These included the Newtsuit , an aluminum atmospheric diving suit for use on drilling rigs and salvage operations. RELATED: Virgin Oceanic’s Voyage to the Bottom of the Sea Deep Rover ’s first assignment was to boost offshore oil exploration and drilling in eastern Canada. Funding came from the provincial government of Newfoundland and Labrador and the oil companies Petro-Canada and Husky Oil. But the collapse of oil prices in the mid-1980s made it uneconomical to operate the submersible. So the rover’s mission broadened to scientific research. Deep Rover ’s Technical Specs The pilot could operate Deep Rover safely for 4 to 6 hours at a depth of 1,000 meters and speeds of up to 1.5 knots (46 meters per minute). The submersible could be tethered to a support ship or move freely on its own. Two deep-cycle, lead-acid battery pods weighing about 170 kilograms apiece provided power. It had a VHF radio and two frequencies of through-water communications, plus tracking beacons. From 1987 to 1989, Deep Rover did a series of dives in Oregon’s Crater Lake, the deepest lake in the United States. During one dive, National Park Service biologist Mark Buktenica [top] collected rock samples. NPS The rover’s four thrusters—two horizontal fixed aft thrusters and two rotating wing thrusters—could be activated in any combination through microswitches built into the armrest. The pilot navigated using a gyro compass, sonar, and depth gauges (both digital and analog). Much to Earle’s delight, Deep Rover had two excellent manipulators, each with four degrees of freedom, thus solving the problem that had started her down this path of invention. The pilot controlled the manipulators with a joystick at the end of each armrest. Sensory feedback systems helped the pilot “feel” the force, motion, and touch. The two arms had wraparound jaws and could lift about 90 kg. If something went wrong, Deep Rover carried five days’ worth of life support stores and had a variety of redundant safety features: oxygen and carbon dioxide monitoring equipment; a halon (breathable) fire extinguisher; a full-face BIBS (built-in breathing system) that tapped into the starboard air bank; and a ground fault-detection system. If needed, the rover could surface quickly by jettisoning equipment, including the battery pods and a 90-kg drop weight in the forward bay. In dire circumstances, the pressure hull (the acrylic bubble, that is) could separate from the frame, taking with it only its oxygen tanks, strobe, through-water communications, and wing thrusters. Deep Rover’s achievements From 1984 to 1992, Deep Rover conducted about 280 dives. It inspected two of the tunnels near Niagara Falls that divert water to the Sir Adam Beck II hydroelectric plant. In California’s Monterey Bay, the rover let researchers film previously unknown deep-sea marine life, which helped establish the Monterey Bay Aquarium Research Institute. At Crater Lake National Park, in Oregon, Deep Rover proved the existence of geothermal vents and bacteria mats, leading to the protection of the site from extractive drilling. Deep Rover was featured in a short film shown at Vancouver’s Expo ’86, the first of several TV and movie appearances. There was Danger Bay . Director James Cameron used an early prototype of the submersible in his 1989 film The Abyss . Deep Rover also made an appearance in Cameron’s 2005 documentary Aliens of the Deep . In 1992, Deep Rover came to the end of its working life. It now resides at Ingenium , Canada’s Museums of Science and Innovation, in Ottawa. For a time, Deep Ocean Engineering continued to develop later generations of the submersible. Eventually, though, uncrewed remotely operated and autonomous underwater vehicles became the norm for deep-sea missions, replacing human pilots with sensors and equipment. New ROVs can dive significantly deeper than human-piloted ones, and new cameras are so good that it feels like you’re there…almost. And yet, humans still long to have the personal experience of exploring the depths of the oceans. Part of a continuing series looking at historical artifacts that embrace the boundless potential of technology. An abridged version of this article appears in the April 2026 print issue as “ All Alone in the Abyss .” References My friends at Ingenium , Canada’s Museums of Science and Innovation, helpfully provided me with background material on why they decided to acquire Deep Rover . They also published a great blog post about the rover. Dirk Rosen , executive vice president of engineering at DEEP, published specifications for Deep Rover in his 1986 IEEE paper “ Design and Application of the Deep Rover Submersible .” Sylvia Earle, known affectionately as “Her Deepness,” has written extensively about the ocean depths. I found her book Sea Change: A Message of the Oceans (G.P. Putnam’s Sons, 1995) to be especially enjoyable.
spectrum.ieee.org
2 months ago
This is a sponsored article brought to you by General Motors. Visit their new Engineering Blog for more insights. Autonomous driving is one of the most demanding problems in physical AI. An automated system must interpret a chaotic, ever-changing world in real time—navigating uncertainty, predicting human behavior, and operating safely across an immense range of environments and edge cases. At General Motors, we approach this problem from a simple premise: while most moments on the road are predictable, the rare, ambiguous, and unexpected events — the long tail — are what ultimately defines whether an autonomous system is safe, reliable, and ready for deployment at scale. (Note: While here we discuss research and emerging technologies to solve the long tail required for full general autonomy, we also discuss our current approach or solving 99% of everyday autonomous driving in a deep dive on Compound AI.) As GM advances toward eyes-off highway driving, and ultimately toward fully autonomous vehicles, solving the long tail becomes the central engineering challenge. It requires developing systems that can be counted on to behave sensibly in the most unexpected conditions. GM is building scalable driving AI to meet that challenge — combining large-scale simulation, reinforcement learning, and foundation-model-based reasoning to train autonomous systems at a scale and speed that would be impossible in the real world alone. Stress-testing for the long tail Long-tail scenarios of autonomous driving come in a few varieties. Some are notable for their rareness. There’s a mattress on the road. A fire hydrant bursts. A massive power outage in San Francisco that disabled traffic lights required driverless vehicles to navigate never-before experienced challenges. These rare system-level interactions, especially in dense urban environments, show how unexpected edge cases can cascade at scale. But long-tail challenges don’t just come in the form of once-in-a-lifetime rarities. They also manifest as everyday scenarios that require characteristically human courtesy or common sense. How do you queue up for a spot without blocking traffic in a crowded parking lot? Or navigate a construction zone, guided by gesturing workers and ad-hoc signs? These are simple challenges for a human driver but require inventive engineering to handle flawlessly with a machine. Autonomous driving scenario demand curve Deploying vision language models One tool GM is developing to tackle these nuanced scenarios is the use of Vision Language Action (VLA) models. Starting with a standard Vision Language Model, which leverages internet-scale knowledge to make sense of images, GM engineers use specialized decoding heads to fine-tune for distinct driving-related tasks. The resulting VLA can make sense of vehicle trajectories and detect 3D objects on top of its general image-recognition capabilities. These tuned models enable a vehicle to recognize that a police officer’s hand gesture overrides a red traffic light or to identify what a “loading zone” at a busy airport terminal might look like. These models can also generate reasoning traces that help engineers and safety operators understand why a maneuver occurred — an important tool for debugging, validation, and trust. Testing hazardous scenarios in high-fidelity simulations The trouble is: driving requires split-second reaction times so any excess latency poses an especially critical problem. To solve this, GM is developing a “Dual Frequency VLA.” This large-scale model runs at a lower frequency to make high-level semantic decisions (“Is that object in the road a branch or a cinder block?”), while a smaller, highly efficient model handles the immediate, high-frequency spatial control (steering and braking). This hybrid approach allows the vehicle to benefit from deep semantic reasoning without sacrificing the split-second reaction times required for safe driving. But dealing with an edge case safely requires that the model not only understand what it is looking at but also understand how to sensibly drive through the challenge it’s identified. For that, there is no substitute for experience. Which is why, each day, we run millions of high-fidelity closed loop simulations , equivalent to tens of thousands of human driving days, compressed into hours of simulation. We can replay actual events, modify real-world data to create new virtual scenarios, or design new ones entirely from scratch. This allows us to regularly test the system against hazardous scenarios that would be nearly impossible to encounter safely in the real world. Synthetic data for the hardest cases Where do these simulated scenarios come from? GM engineers employ a whole host of AI technologies to produce novel training data that can model extreme situations while remaining grounded in reality. GM’s “Seed-to-Seed Translation” research , for instance, leverages diffusion models to transform existing real-world data, allowing a researcher to turn a clear-day recording into a rainy or foggy night while perfectly preserving the scene’s geometry. The result? A “domain change”—clear becomes rainy, but everything else remains the same. In addition, our GM World diffusion-based simulator allows us to synthesize entirely new traffic scenarios using natural language and spatial bounding boxes. We can summon entirely new scenarios with different weather patterns. We can also take an existing road scene and add challenging new elements, such as a vehicle cutting into our path. High-fidelity simulation isn’t always the best tool for every learning task. Photorealistic rendering is essential for training perception systems to recognize objects in varied conditions. But when the goal is teaching decision-making and tactical planning—when to merge, or how to navigate an intersection—the computationally expensive details matter less than spatial relationships and traffic dynamics. AI systems may need billions or even trillions of lightweight examples to support reinforcement learning, where models learn the rules of sensible driving through rapid trial and error rather than relying on imitation alone. To this end, General Motors has developed a proprietary, multi-agent reinforcement learning simulator, GM Gym, to serve as a closed-loop simulation environment that can both simulate high-fidelity sensor data, and model thousands of drivers per second in an abstract environment known as “Boxworld.” By focusing on essentials like spatial positioning, velocity and rules of the road while stripping away details like puddles and potholes, Boxworld creates a high-speed training environment for reinforcement learning models at incredible speeds, operating 50,000 times faster than real-time and simulating 1,000 km of driving per second of GPU time. It’s a method that allows us to not just imitate humans, but to develop driving models that have verifiable objective outcomes, like safety and progress. From abstract policy to real-world driving Of course, the route from your home to your office does not run through Boxworld. It passes through a world of asphalt, shadows, and weather. So, to bring that conceptual expertise into the real world, GM is one of the first to employ a technique called “On Policy Distillation,” where engineers run their simulator in both modes simultaneously: the abstract, high-speed Boxworld and the high-fidelity sensor mode. Here, the reinforcement learning model—which has practiced countless abstract miles to develop a perfect “policy,” or driving strategy—acts as a teacher. It guides its “student,” the model that will eventually live in the car. This transfer of wisdom is incredibly efficient; just 30 minutes of distillation can capture the equivalent of 12 hours of raw reinforcement learning, allowing the real-world model to rapidly inherit the safety instincts its cousin painstakingly honed in simulation. Designing failures before they happen Simulation isn’t just about training the model to drive well, though; it’s also about trying to make it fail. To rigorously stress-test the system, GM utilizes a differentiable pipeline called SHIFT3D . Instead of just recreating the world, SHIFT3D actively modifies it to create “adversarial” objects designed to trick the perception system. The pipeline takes a standard object, like a sedan, and subtly morphs its shape and pose until it becomes a “challenging”, fun-house version that is harder for the AI to detect. Optimizing these failure modes is what allows engineers to preemptively discover safety risks before they ever appear on the road. Iteratively retraining the model on these generated “hard” objects has been shown to reduce near-miss collisions by over 30%, closing the safety gap on edge cases that might otherwise be missed. Even with advanced simulation and adversarial testing, a truly robust system must know its own limits. To enable safety in the face of the unknown, GM researchers add a specialized “Epistemic uncertainty head” to their models. This architectural addition allows the AI to distinguish between standard noise and genuine confusion. When the model encounters a scenario it doesn’t understand—a true “long tail” event—it signals high epistemic uncertainty. This acts as a principled proxy for data mining, automatically flagging the most confusing and high-value examples for engineers to analyze and add to the training set. This rigorous, multi-faceted approach—from “Boxworld” strategy to adversarial stress-testing—is General Motors’ proposed framework for solving the final 1% of autonomy. And while it serves as the foundation for future development, it also surfaces new research challenges that engineers must address. How do we balance the essentially unlimited data from Reinforcement Learning with the finite but richer data we get from real-world driving? How close can we get to full, human-like driving by writing down a reward function? Can we go beyond domain change to generate completely new scenarios with novel objects? Solving the long tail at scale Working toward solving the long tail of autonomy is not about a single model or technique. It requires an ecosystem — one that combines high-fidelity simulation with abstract learning environments, reinforcement learning with imitation, and semantic reasoning with split-second control. This approach does more than improve performance on average cases. It is designed to surface the rare, ambiguous, and difficult scenarios that determine whether autonomy is truly ready to operate without human supervision. There are still open research questions. How human-like can a driving policy become when optimized through reward functions? How do we best combine unlimited simulated experience with the richer priors embedded in real human driving? And how far can generative world models take us in creating meaningful, safety-critical edge cases? Answering these questions is central to the future of autonomous driving. At GM, we are building the tools, infrastructure, and research culture needed to address them — not at small scale, but at the scale required for real vehicles, real customers, and real roads.
spectrum.ieee.org
2 months ago
A growing number of Nigerian companies are turning to kit-based assembly to bring electric vehicles to market in Africa. Lagos-based Saglev Micromobility Nigeria recently partnered with Dongfeng Motor Corp. , in Wuhan, China, to assemble 18-seat electric passenger vans from imported kits. Kit-based assembly allows Nigerian firms to reduce costs, create jobs, and develop local technical expertise—key steps toward expanding EV access. Fully assembled and imported EVs face high tariffs that put them out of reach for many African consumers, whereas kit-based approaches make electric mobility more affordable today. Saglev’s initiative reflects a broader trend: CIG Motors , NEV Electric , and regional players in Côte D’Ivoire, Ghana, and Kenya are also leveraging imported kits to build local EV ecosystems, signaling that parts of West Africa are intent on catching up with global electrification efforts. Expanding the Local EV Ecosystem CIG Motors operates a kit-assembly plant in Lagos producing vehicles from Chinese automakers GAC Motor and Wuling Motors . These vehicles include the Wuling Bingo , a compact five-door electric hatchback, and the Hongguang Mini EV Macaron , a microcar with roughly 200 kilometers of range aimed at ride-share operators looking for ultralow-cost urban transport. NEV Electric focuses on electric buses and three-wheelers for urban transit and last-mile delivery. Saglev’s CEO, Olu Faleye, emphasizes that Nigeria’s EV transition addresses both practical economic needs in addition to environmental goals. Beyond passenger transport, electric vehicles could help reduce one of Nigeria’s persistent agricultural challenges: postharvest spoilage. Nigeria loses an estimated 30 million to 40 million tonnes of food annually because of weak logistics and limited refrigeration infrastructure, according to the Organization for Technology Advancement of Cold Chain in West Africa . Electric vans, minitrucks, and three-wheel cargo vehicles could help close this gap because their batteries can power refrigeration systems during transport without relying on costly diesel fuel. As EV adoption grows and charging infrastructure expands, temperature-controlled transport could become more affordable, reducing spoilage, improving farmer incomes, and helping stabilize food supplies, the organization says. “I don’t believe that the promised land is making a fully built EV on the ground here.” –Olu Faleye, Saglev CEO Beyond Nigeria, Mombasa, Kenya–based Associated Vehicle Assemblers has begun making electric taxis and minibuses from imported kits, and Ghana’s government is spurring kit-car assembly there under its national Automotive Development Plan . In Ghana, assemblers benefit from import-duty exemptions on kits and equipment, corporate tax breaks, and access to industrial infrastructure. Saglev is already availing itself of those benefits, at its kit-assembly plant in Accra, Ghana. The company says it also plans to expand its assembly operations to Côte D’Ivoire. Infrastructure Challenges and Workarounds Despite these signs that West Africa’s EV ecosystem is gaining traction, limited grid reliability and sparse public charging infrastructure remain major barriers to widespread EV adoption. Urban households in Nigeria experience roughly six or seven blackouts per week , each lasting about 12 hours, according to Nigeria’s National Bureau of Statistics . That’s more downtime each day than the average U.S. household experiences in a year. More than 40 percent of households rely on generators , which supply about 44 percent of residential electricity, according to research by Stears and Sterling Bank . Many early EV adopters therefore charge vehicles using gasoline or diesel generators. Faleye notes that Nigerians have long relied on such workarounds and expects fossil fuels to remain part of the EV charging equation for the foreseeable future—at least until falling costs for solar panels and battery storage make cleaner charging viable. He acknowledges that charging EVs using hydrocarbons is fraught from an environmental perspective, but he points out that the practice at least brings other benefits of EVs, including lower maintenance costs and the EVs’ synergies with refrigeration and transportation logistics . And he points to a 2020 peer-reviewed study in the journal Environmental and Climate Technologies that compared the overall efficiency of internal combustion vehicles and electric vehicles across the full well-to-wheel energy chain. The study’s conclusion: Even after accounting for conversion losses, generating electricity with a diesel or gasoline generator to power an electric vehicle can remain just as efficient overall as burning the same fuel directly in a vehicle’s internal combustion engine. Workers at Saglev’s Lagos, Nigeria, EV assembly plant put the finishing touches on partially assembled vehicle kits imported from China. Saglev Scalable EV Adoption in Nigeria The approach taken by Saglev and other Nigerian kit-car builders shows how local assembly can advance EV adoption even where infrastructure remains unreliable. By starting with kits, companies can deploy practical electric mobility solutions now while building the supply chains and technical expertise needed for more resource-intensive localized production. Still, when asked whether Saglev plans to eventually move beyond kit assembly to independent design and manufacturing of EVs, Faleye calls such a move impractical. “I don’t believe that the promised land is making a fully built EV on the ground here,” he says. “For me to do efficient vehicle manufacturing, I’d need a lot of robotics and 3D printing . That expense is unnecessary—it would just increase costs and make EVs more expensive.” In a country where electricity can disappear for days, Nigeria’s kit-based EV strategy highlights a practical truth: Incremental progress and ingenuity may matter more than perfect infrastructure. For Saglev, every kit-based vehicle rolling off the line is not just a van or bus—it’s a step toward an EV ecosystem that works for Nigeria’s realities today.
spectrum.ieee.org
2 months ago
As drivers increasingly ditch gas-guzzling cars for electric vehicles, a persistent problem remains for pedestrians: hearing them coming. To address that risk, regulators in Australia , Europe , the United States , and other regions now require EVs to emit artificial warning sounds at low speeds, forcing some automakers to recall their cars for failing to meet safety standards. What might seem like a simple safety measure has become a complex web of engineering, involving acoustics, signal design, human perception, and regulatory compliance. Michael Roan Michael Roan is a professor of mechanical engineering at Penn State University. His research includes how artificial vehicle sounds are perceived, particularly by visually impaired pedestrians. Automakers including BMW , Hyundai , Mercedes-Benz and others have enlisted sound designers, acoustics specialists, and even musicians to craft audio signatures that satisfy safety mandates while reinforcing brand identity. But designing those signals means navigating strict technical requirements and the messy realities of urban soundscapes. And it’s still not entirely clear how well those warning signals actually work in real-world conditions. To learn more about the many challenges that automakers confront, IEEE Spectrum spoke with Michael Roan , an acoustics professor at Penn State University who has researched how artificial vehicle sounds are perceived, particularly by visually impaired pedestrians. What challenges do automakers encounter when creating sounds for EVs? Michael Roan: There’s a lot of challenges. A speaker has a certain frequency response. When you put it somewhere, you face the possibility that there’ll be dead zones. If I transmit some sound and it’s in an enclosed area, that sound doesn’t propagate straight out to the listener. It bounces off the road and inside the enclosure that it’s in, whether that’s the wheel well or under the hood. The signals that they use are very tonal, and so that creates the possibility of destructive interference and constructive interference. That means you could have spots where it’s really loud, and then spots where you can’t hear it at all. Trying to make a source that is even around the exterior of the vehicle is really challenging. When automakers are creating EV sounds, what are they trying to optimize for? And how do different priorities clash? Roan: It’s really about safety, then brand. And then, of course, cost—how much is it gonna cost me to put this system in 100,000 vehicles, and have it be reliable, and not cost a fortune. The clash is figuring out how to satisfy the safety regulations without being annoying. I’m sure you’ve heard plenty of electric vehicles creeping up on you while you’re walking on the sidewalk, and it plays that whirring spaceship sound. It’s shrill, hard-edged, and very tonal. That’s satisfying the regulations, but it’s not aesthetically pleasing. So how can you do both? That’s where you have to mix in a little artistry to make sure that’s happening. There’s perceptible, but then there’s annoying. In Europe, people are particularly sensitive to noise pollution. If one of these cars is sitting there in an intersection, it’s okay. But what happens when there’s 50 sitting in an intersection? Is that going to destroy the acoustic environment for people living in that area? Balancing all that is a really complex issue. Real-World EV Sound Challenges How do EV sounds typically hold up in real world conditions with traffic, construction, and other ambient noise? Roan: I don’t think anyone has answered that yet. I bet nobody would say that ‘if you satisfy the regulation, then in a real world situation, you’ll be X amount of safe.’ That doesn’t exist, because nobody really knows. It hasn’t been tested in a real world environment where you put out listeners in a real intersection in a city like New York City. But that’s just a gut feeling based on experiences that hasn’t been proven out. Is there interest in standardizing EV sounds, where one sound could be a one size fits all type of solution? Roan: I don’t think the car makers would want to do it. They really like to separate themselves, to have some sort of identity that’s different from the other car makers. That’s what makes them special. What would it take to create a speaker that is loud enough and meets regulatory standards without being annoying? Roan: There’s been a lot of things looked at. For instance, we played around with putting a piezoelectric actuator on the hood and using the hood as a source of sound. One [option] that would be really great is you put an array of speakers on the bumper in the front and the back and beamformed it. That way, it focuses on the sidewalk where people would only hear it as the car went by in a very narrow spatial range. But you need this array of speakers as well as signal processing to beam that sound where you need it to go. That would be ideal for lowering noise pollution, but it’s very expensive. Do you think the current EV stand regulations meaningfully improve safety, or are they mostly compliance driven? Roan: I think they improve safety. When we did our testing on a Chevrolet Bolt car, we did a control set where there was no added sound and tested them against different additive sounds. For the no sound case, the probability detection was significantly lower. It needs to be there. I think it’s a great start. But there’s not a lot of research going on in EV sound safety anymore. In a sense, the [U.S.] government’s like, ‘Well, we’ve got to figure it out, have a nice day.’ But as more EVs roll out, maybe we need to go to Norway where they have a huge amount of EVs on the road to see how their safety is going. I don’t know if anybody’s even looking at that, to tell you the truth. A version of this interview appears in the April 2026 print issue as “Michael Roan.”
spectrum.ieee.org
2 months ago
Raquel Urtasun has spent 16 years in the self-driving space , long enough to navigate every metaphorical glorious hill and plunging valley . She took the trip from the early “pipe dream” dismissals to the “we’re this close” certainty and back again. The industry is now riding a new wave of optimism and investment, including at Waabi Innovation Inc. , the autonomous trucking company that Urtasun founded in 2021. The Spanish-Canadian professor at the University of Toronto , and former chief scientist of Uber’s Advanced Technologies Group, has helped make Waabi a key player. Beginning in fall 2023, the Toronto-based startup has been running geofenced cargo routes from Dallas to Houston in a fleet of retrofitted Peterbilt semis, navigating even residential streets in loaded, 36,000-kilogram ( 80,000-pound ) behemoths with a human “safety observer” on board. In October, the company reached a milestone by integrating its “Waabi Driver” physical-AI system in Volvo’s new VNL Autonomous truck, which the Swedish automaker is building in Virginia. That self-driving solution uses Nvidia’s Drive AGX Thor , an AI-based platform for autonomous and software-defined vehicles. In January, the Toronto-based startup raised $750 million in its latest funding round to accelerate commercial development in autonomous trucking and expand its system into the fiercely competitive robotaxi space. Backers include Khosla Ventures , Nvidia , and Volvo . Urtasun says the Waabi Driver can scale across a full range of vehicles, geographies, and environments— although snowstorms can still create a no-go zone for now. It’s powered by what Urtasun calls the industry’s most advanced neural simulator. The verifiable, end-to-end AI model will be a “shared brain” that partners can transplant into cars, trucks, and pretty much anything on wheels . The idea is to grab a chunk of a global autonomous trucking business that McKinsey estimates could be worth more than $600 billion a year by 2035, with autonomous haulers responsible for 15 percent of total U.S. trucking miles as early as 2030. Backed by an additional $250 million from Uber , Waabi plans to deploy at least 25,000 autonomous taxis through Uber’s ride-hailing service, whose world-dominating reach encompasses 70 countries, about 15,000 cities, and more than 200 million monthly users. Urtasun spoke with IEEE Spectrum about how Waabi is counting on sensors and simulation to prove real-world safety, and why the move to autonomy is a moral imperative that outweighs the disruption for human drivers— whether they’re driving trucks or family sedans . Our conversation was edited for length and clarity. The Shift to Next-Gen Autonomous Vehicles IEEE Spectrum : Until quite recently, autonomous tech seemed to have hit a wall, at least in the public’s mind. Now investors are flooding the zone again, and companies are all in. What happened? Raquel Urtasun: There were a lot of empty promises, or [people] not realizing the complexity of the problem. There was a realization that actually, this problem is harder than people anticipated. It’s also because of the type of technology that was developed at the time, what we call “AV 1.0.” These are hand-engineered systems that need to be brute-forced by humans. You need lots of capital and a massive amount of miles on the road just to get to the first deployment. What you see with the next generation— AV 2.0 and systems that can reason—is that you finally have a solution that scales. When we started the company, this was a very contrarian view. But today, the breakthroughs in AI have made it clear that this is the next big revolution. It’s not just about more compute; it’s about building a brain that can generalize. That is the “aha moment” the industry is having now. Even for someone who believes in the tech, seeing a driverless semi-trailer in your rearview mirror might be unsettling. Now you’ve integrated your tech into the aerodynamic, diesel-powered Volvo VNL Autonomous truck. How do you convince regulators and the public that these trucks belong on the street? Urtasun: Safety, when you think about carrying 80,000 pounds on this massive rig, is definitely top of mind. We believe the only way to do this safely is with a redundant platform that is fully developed and validated by the OEM, not with a retrofit. The OEM does a special type of truck that has all the redundant steering, power, and braking, so that no matter what happens, there is always a way we can interface and activate that truck in a safe manner. Then we are responsible for the sensors, the compute, and obviously the brain that drives those trucks. AI’s Impact on Trucking Jobs One of the biggest points of contention is the displacement of human drivers. As AI disrupts a range of workplaces, how do you respond to people who say this will eliminate good-paying blue-collar jobs? Urtasun: The way we see this is that everybody who’s a truck driver today, and wants to retire as a truck driver, will be able to do so. This is physical AI; this is not like the digital world, where suddenly you can switch immediately to this technology. That adoption and scaling is going to take time. There will also be many jobs created with this technology; remote operations, terminal operations, and other things. You have time to change the form of labor of being on the road, which is for weeks at a time—and it’s a really difficult and dehumanized job, let’s be honest—to something you can do locally. There was an interesting [U.S.] Department of Transportation study that showed because of this gradual adoption, there will be more jobs created than actually removed. You’ve spoken about a personal motivation behind this. Why do you believe the advantages of autonomy outweigh any growing pains, including the potential for unexpected accidents or even deaths? Urtasun: There are 2 million deaths on the road globally per year, and nobody’s questioning that. That’s the status quo. If you think the machines have to be perfect to deploy, you are actually sacrificing many humans along the way that you could have saved. Human error in accidents is between 90 percent and 96 percent . Those could be preventable accidents. Some accidents will always be unavoidable; a tire could blow for a machine the same as it could for a human. But the important comparison is how much safer we are. This technology is the answer to many, many things. Most of the industry is focused on “hub-to-hub” highway driving. But you’ve argued that Waabi’s AI can handle the complexity of local streets. Urtasun: The rest of the industry has gone with this business model, where you need hubs next to the highway. This adds a lot of friction and cost. Thanks to our verifiable end-to-end AI system, we can drive in surface [local] streets. We can do unprotected lefts, traffic lights, and tight turns. These core capabilities enable us to drive all the way to the end customer. We are already hauling commercial loads for customers like Samsung through our Uber Freight partnership. You’ve mentioned that Waabi doesn’t like to talk about “number of miles” driven as a metric. For an engineering audience, that sounds counterintuitive. How does your “simulation-first” approach replace the need for real-world road time? Urtasun: In the industry, miles have been used as a proxy for advancement. How many miles does Tesla need to drive to see any of these situations? But we are a simulation-first company. Waabi World can simulate all the sensors, the behaviors of humans, everything. It is the only simulator where you can mathematically prove that testing and driving in simulation is the same as driving in the real world. You can expose the system to billions of simulations in the cloud. This is what allows us to be so capital efficient and fast. Verifiable AI vs. Black Box Systems What is the difference between your “interpretable” AI and the “black box” systems we see elsewhere? Urtasun: We’ve seen an evolution on passenger cars for level - 2+ systems to end-to-end, black box architectures. But those are not verifiable. You cannot validate and verify those systems, which is a massive problem when you think about regulators and OEMs trusting that technology. What Waabi has built is end-to-end, but fully verifiable. The system is forced to interpret what it is perceiving and use those interpretations for reasoning, so that it can understand the consequences of every action. It is much more akin to how our brain actually works; your “Type 2” thinking, where you start thinking about cause and effect and consequences, and then you typically do a much better choice in your maneuver. Tesla is famously, and controversially, relying on camera data almost exclusively to run and improve its self-driving systems. You’re not a fan of that approach? Urtasun: We use multiple sensors: lidar, camera, and radar. That’s very important, because failure modes of those sensors are very different, and they’re very complementary. We don’t compromise safety to reduce the bill-of-materials cost today. Those (passenger car) l evel- 2+ systems are not architected for level 4 , where there’s no human on board. People don’t necessarily realize there is a huge difference in terms of the bar when there is no human to rely on. It’s not, “Well, if I don’t have a lot of system interventions, I’m almost there.” That’s not a metric. We are native level 4. We decide which areas the system can drive in, and in what conditions. We are building technology that can drive different form factors—trucks or robotaxis—with the same brain. Editor’s note: This article was updated on 13 March to correct an error in the original post. Contrary to what was stated in the original post, the trucks being driven from Dallas to Houston do have a human observer on board.
spectrum.ieee.org
2 months ago
This past January, Donut Lab sparked a worldwide battery controversy at CES 2026 by announcing it had developed a solid-state battery for a forthcoming Verge electric motorcycle. The Finland-based Donut claimed a commercial breakthrough that has eluded the world’s largest battery companies or startups, whether China’s CATL , BYD , Factorial Energy , or QuantumScape . Skeptics have looked to blow a hole in Donut’s claims ever since. Yang Hongxin , chairman and chief executive of Chinese battery maker SVOLT Energy , did not mince words . “That battery doesn’t exist in the world,” Hongxin told local media after the company’s Battery Day in January. “All the parameters are contradictory.…Any technician with basic knowledge would recognize it as a scam.” Donut, which previously garnered attention for its hubless in-wheel electric motors , has met the doubt with defiance. In an interview with IEEE Spectrum , Ville Piippo , Donut’s co-founder and CTO, acknowledged natural skepticism toward a company with no proven track record in batteries. “If the world is pouring billions and billions of dollars into solid-state, why haven’t they figured this out?” Piippo asked rhetorically. “The answer is the same as for our motors, that we are doing things a different way, and the rest of the world has focused on the wrong thing.” The company has launched a website, idonutbelieve , to directly address the backlash and company critics. To help back its claims, the company began posting a video series on 23 February that highlights third-party testing of its technology by the state-owned VTT Technical Research Centre of Finland. Extraordinary Claims Require Extraordinary Evidence The claims are dramatic. Donut says its solid-state battery can charge to 80 percent of capacity in as little as five minutes , endure 100,000 cycles, and deliver 400 watt-hours per kilogram of energy density, versus the 200-300 Wh/kg of typical lithium-ion cells. With no liquid electrolyte, solid-state batteries should be virtually immune to thermal runaway. Donut is a spin-off of Estonia’s Verge Motorcycles, whose Verge TS Pro was expected to begin shipping to customers in the first quarter of 2026, with up to 600 kilometers (370 miles) of riding range from its larger, optional 33.3 kilowatt-hour pack. Independent testing by VTT in recent months is backing some company claims, including for blazing charging speeds. But Eric Wachsman , a solid-state battery expert and director of the Maryland Energy Innovation Institute, says the tests raise as many questions as they answer. To name two, the company has still not revealed the chemistry inside the cells, and the test did not weigh the cells to determine energy-density figures. “They’re quoting 400 watt-hours per kilogram, but nothing here is telling me what the masses or volumes are,” Wachsman says. VTT tested a Donut pouch cell with 24 amp-hours, a nominal voltage of 3.6 and nominal energy of 94 watt-hours. Cells were charged at both 5C and 11C rates. These designations are based on a 1C rate, which means the battery can be fully charged in one hour. 5C means the battery can be charged in 1/5 hour, or 12 minutes. Tests were conducted with passive aluminum cooling plates, both single- and double-sided, to simulate thermal management in an EV. The 5C test zapped the cell with 130 amps at 4.3 volts and achieved an 80 percent charge in 9.5 minutes, and 100 percent in just over 12 minutes. Peak temperature reached 47 ºC (116.6 ºF). The same cell was then charged at an extreme speed of 11C, which showed a 0-to-80-percent charge in 4.6 minutes—which validated Donut’s claim that the battery can be charged from 0 to 80 percent within 5 minutes. The VTT test further indicated a 0 to 100 percent charge at 11C in just over 7 minutes. Temperatures reached 63 ºC (145 ºF). A subsequent discharge and charge left the battery with up to 99.6 percent of its original capacity available. Seven Charge-Discharge Cycles Tested, 99,993 To Go Wachsman notes the cells were only tested over seven cycles. That doesn’t begin to prove the cells could last hundreds or thousands of cycles—let alone the 100,000 cycles that Donut claims, which is exponentially more than the 2,000-3,000 full charge-discharge cycles typical of the lithium-ion batteries now used in EVs. A second test measured the cell’s performance under high-temperature conditions . Those tests raised more red flags with Wachsman, including graphs that suggested an unimpressive round-trip-efficiency (RTE), or the ratio of usable energy retrieved during discharge, versus the energy inputs during charging. The tests showed roughly 90 percent RTE, including a cycle that delivered 99.97 watt-hours of charge, and 90.36 watt-hours of discharge energy. Designers of solid-state batteries are aiming for RTE’s of 98 percent or higher, due in part to their inherently lower internal resistance. And while Donut has claimed its batteries don’t exhibit much volume change during cycles, or require extensive external compression to control volumes—which inevitably adds mass to packs—Wachsman said images from the test seem to show the battery had visibly swelled during charging. VTT itself notes that one pouch cell lost its vacuum seal during testing. “Swelling is a problem with batteries in general, after hundreds or thousands of cycles, and this was after four cycles,” Wachsman notes. Experts also point to the usual conundrums that bedevil battery developers. The tests don’t demonstrate how the cells might perform at the pack level. They don’t prove that Donut—which has yet to reveal where cells or packs will be manufactured—can make the leap from the lab to manufacturing defect-free batteries at scale. Beyond chemistries, those are roadblocks that have prevented even the CATLs, BYDs, and Samsungs of the world from commercializing solid-state. Marko Lehtimäki , cofounder of Donut and Verge, says the proof is in the product, though the release date of the Verge TS Pro may now be pushed back. Responding to reports of delays, Lehtimäki promised that some early customers will still see their motorcycles by the end of March. But customers who order bikes after early February won’t take delivery until the fourth quarter. The company’s site features a countdown clock to the next test-result reveal, which on my 2 March visit read 6 days, 12 hours, 32 minutes, and counting. Yet until the Verge TS Pro ends up in customers hands, and people can test it or start tearing these batteries down, the clock won’t stop ticking on this controversy. This article appears in the May 2026 print issue as “ Donut Lab Says Its Solid-State Battery Is Legitimate .”
spectrum.ieee.org
3 months ago
Most 3D printers are designed to produce plastic parts, such as prototypes, housings for electronics , or decorative objects. Building a working electric machine is far more complicated. Unlike a typical plastic print, devices like motors need different regions to do different jobs: Some conduct electricity, others insulate it, some generate or guide magnetic fields, and others provide structural support or flexibility. A research team at MIT is trying to bring 3D printing and functional hardware design together. In a paper published last month in Virtual and Physical Prototyping , the group introduced a multimaterial 3D-printing system capable of producing a working electric linear motor in about three hours. The platform processes five functional materials used in the printed motor , including conductors, magnetic structures, and flexible components. The motor required about US $0.50 in raw materials, with the researchers positioning the platform as a way to make hardware engineering cheaper, faster, more local, and less vulnerable to supply chain disruptions. Conventional electric motors are typically assembled from separately manufactured components using multiple fabrication steps. The MIT system, however, prints the functional structures directly in one build process, with only a single postprint step to magnetize the motor’s hard magnetic parts. The researchers started with a demo of a working linear motor—an actuator that produces smooth motion in a straight line. Eventually, they hope to scale the concept to more complex rotating motors used in vehicles and other heavy-duty applications. How Multimaterial 3D Printing Works 3D printing has evolved significantly from its origins in rapid prototyping, but most printers operating today are still single-material, single-nozzle machines optimized for plastics. Even systems marketed as “multimaterial” often mean “multicolor,” using the same underlying polymer rather than fundamentally different materials. Many extrusion-based printers swap between two similar feedstocks, such as filament, rather than combining distinct functional materials. “Very few applications can be satisfied with just one material,” says Luis Fernando Velásquez-García , principal research scientist at MIT Microsystems Technology Laboratories, who led the study. “If you want to make hardware that actually does something well, it usually requires different materials.” Velásquez-García argues multimaterial extrusion is the best approach for producing functional hardware. In MIT’s prototype, the printer can switch among four tools to handle feedstocks with very different properties: a heater for curing ink, a filament extruder, a custom ink extruder, and a modified pellet extruder. The pellet-based extrusion tool is particularly useful because it allows higher magnetic particle concentrations. In standard filament printing, material is fed as a thin plastic strand, which limits how many particles can be mixed into the polymer before the filament becomes brittle. Pellet extrusion instead feeds the printer small plastic pellets, allowing much higher particle loading and improving the magnetic performance of printed components. Velásquez-García emphasized the use of capable materials, not just printable ones. Many limitations of printed devices stem from the underlying materials used to fabricate them. If the base material’s fundamental properties fall short, performance will suffer. “You shouldn’t make any compromises in materials and performance. If you need something with optical clarity, for example, it has to be very clear or it won’t work,” Velásquez-García says. “The goal...should be to deliver hardware that does what people want. And if the products can be made with printing, that’s great.” He adds that, done properly, “multimaterial 3D printing is actually a win-win situation.” Although material variety matters a lot, so does process compatibility. For example, conductive inks often need specific curing conditions to avoid damaging the insulating materials. And even if each material can be printed, the device works only if every layer aligns precisely. According to MIT, the team addressed this with a strategic sensor setup and control system so the robotic arms can interchange tools consistently and predictably. MIT’s multimaterial system can switch among four tools: [clockwise from top left] a filament extruder, pellet extruder with custom 3D-printed parts, heater for ink curing, and custom-built ink extruder. Jorge Cañada, Zoey Bigelow and Luis Fernando Velásquez-García A 3D-Printed Linear Motor For its demonstration, the MIT team focused on a linear motor, a device commonly used in high-precision systems like chip wafer manufacturing , pick-and-place robotics, medical imaging, and conveyance systems. These motors also provide a useful proving ground for printed electromagnetics. Velásquez-García said the prototype system was built from a mix of off-the-shelf components and custom parts, and it cost “on the order of a few thousand dollars.” The platform prints linear motors using five functional material classes: dielectric, electrically conductive, soft and hard magnetic, and flexible. The researchers reported that the printed motor performed comparably or better than devices built through conventional multistep fabrication methods, while requiring only a single step after the print—to magnetize the motor. They also said it could produce more actuation than typical linear systems that generate force through hydraulic amplifiers. Still, this shouldn’t be mistaken for a printed EV drivetrain. The team’s next target is a more complex class of device: rotating motors. Those systems place harsher constraints around coil density, thermal management, and mechanical durability. Electric vehicles are one example of where such motors are used, but Velásquez-García stressed that the research is still far from that scale. “There’s a long way between what we have and a 3D-printed engine in an electric car,” Velásquez-García says. “We’re far from that because we would need to make something that rotates and can deal with the temperature, load, and other things. So I think it’s exciting, but I don’t want to oversell it. It’s research and there are still a number of things that we can do, but I think it’s exciting because it’s the first of its kind.” The team’s next goals include incorporating magnetization directly into the printing process, expanding the system with additional tools, and eventually demonstrating fully 3D-printed rotary motors—steps toward producing more complex electronic systems on a single platform. Velásquez-García says the full system could allow engineers to combine dissimilar materials in a single print to build functional electromechanical designs remotely, far from manufacturing hubs. The long-term value is that a repair team, a remote station, or a small manufacturer could eventually fabricate specialized components without waiting on global logistics.
spectrum.ieee.org
3 months ago
At first glance, the Aria EV doesn’t look much different from any other student-built electric prototype—no different from the battery-powered cars built by engineering students from dozens of universities every year. Beneath its panels, however, is a challenge to the modern auto industry: What if electric vehicles were designed to be repaired by their owners? The Aria project began in 2024, when roughly 20 students assembled at Eindhoven University of Technology in the Netherlands under the university’s Ecomotive team structure, which operates like a small startup. Students apply, are selected, and spend a year developing a vehicle in a setting meant to mirror industry practice. The goal, says team spokesperson Sarp Gurel , “was to make the car as accessible and repairable as possible.” Gurel, who graduated last July with a bachelor’s degree in industrial engineering and is currently working toward a master’s degree at Eindhoven, says the Aria EV is not yet road legal. Its purpose is to demonstrate that repairability can be embedded into EV architecture from the outset. With that objective in mind, the team focused first on the most challenging and expensive component in almost any EV: the battery. Modular Battery Design in EVs Aria’s total battery capacity is 13 kilowatt-hours, which is far below the 50- to 80-kWh packs common in mass-market electric sedans and SUVs. The scale is closer to that of a lightweight urban vehicle or neighborhood EV, which is more appropriate for a student-built prototype focused on concept validation rather than long-range highway travel. What distinguishes Aria is not the battery’s size, but its structure. Rather than housing the 13 kWh in a single sealed pack, the team divided the total capacity into six smaller modules. Each module weighs about 12 kilograms—much easier to handle than the 400 kg or more that’s typical of a conventional EV’s monolithic battery pack. This makes it feasible for a single person to remove, swap, and replace modules. The modules sit in reinforced compartments beneath the vehicle floor and are secured using a bottom-latch system. When the vehicle is fully powered down, a latch can be made to mechanically release a module. Integrated interlocks isolate the high-voltage connection before a module can be lowered. This combination of hardware and software ensures that component-level replacement is straightforward and relatively safe, bringing the idea of “repairability by design” into a tangible, hands-on form. Even with this careful design, modular batteries introduce technical considerations that must be managed, particularly when integrating different modules over the vehicle’s lifespan. Joe Borgerson , a laboratory research operations coordinator at Ohio State University ’s Center for Automotive Research , in Columbus, notes one complication: Mixing new and aged battery modules can create challenges. Borgerson has spent the past three years designing and building a battery pack from scratch as part of the U.S. Department of Energy ’s Battery Workforce Challenge . “Our team is integrating a student-designed pack into a Stellantis vehicle platform,” he says, “ which has given me deep exposure to both automaker design philosophy and high-voltage EV architecture, .” To complement their car’s hardware, the Aria team developed a diagnostic app that can be accessed via a dedicated USB-C port. When the user connects their smartphone, the app presents a 3D visualization on the phone screen that points out faults, locates problems, identifies the necessary tools to fix them, and provides step-by-step repair instructions. The tools themselves are stored in the vehicle. The system aims to reduce as many barriers as possible for users to maintain and extend a vehicle’s service life. Students at Eindhoven University of Technology unveiled their Aria EV prototype in November. Sarp Gürel Challenges of EV Modularity While Aria prioritizes modularity, the broader EV industry trend is toward integrated, interdependent systems that simplify manufacturing processes and cut costs. This trend is true for the structural battery packs for EVs as well. Unlike mainstream EVs, Aria treats energy storage as a replaceable subsystem. Whether it scales economically and structurally to larger, highway-capable EVs remains an open question. But designing a vehicle for repairability involves trade-offs that ripple across every system in the car. Borgerson says that dividing systems into removable units adds interfaces—mechanical fasteners, electrical connectors, seals, and safety interlocks. Each interface must survive vibration, temperature swings, and crash forces. More interfaces can mean added mass and complexity compared with tightly integrated battery structures. And these components take up space that would otherwise be used for energy storage. Matilde D’Arpino , an assistant professor of mechanical and aerospace engineering at Ohio State whose research focuses on electrified power trains and advanced vehicle architectures, notes that EV batteries are already modular internally—cells form modules, and modules form packs—but making modules externally replaceable changes validation requirements. High-voltage isolation, thermal performance, and crash integrity must remain robust even when energy storage is divided into removable segments. In other words, what seems like a simple way to make batteries user-friendly actually cascades into system-level design decisions influencing safety, thermal management, and vehicle structure. Impact of Right-to-Repair Laws Right-to-repair legislation in Europe and the United States could push automakers to reconsider sealed architectures for batteries and other components. Economic incentives could also emerge from fleet operators or long-term owners who benefit from replacing a fraction of a battery system rather than an entire pack. But a dopting this approach would require changes across supply chains, certification processes, and service models. The Aria prototype isn’t ready to go toe-to-toe with production EVs, but it demonstrates some proof-of-concept ideas about repairability. Sarp Gürel Consumer expectations are also shaping the boundaries of what designs like Aria’s can become. In the mainstream market, buyers consistently prioritize longer driving range and lower sticker prices—two factors that have defined competition among models such as the Chevrolet Bolt EV , the Hyundai Ioniq 5 , and the the Tesla Model 3 . Range anxiety remains a powerful psychological factor, even as charging infrastructure expands, and price sensitivity has intensified as government incentives fluctuate. Designing for modularity and repairability, as Aria does, must ultimately contend with these consumer priorities. Any added cost, weight, or complexity must be weighed against a market that still rewards vehicles that go farther for less money. Ultimately, however, Aria inserts a different priority into the equation: repair as a core design requirement. Whether that priority becomes mainstream will depend less on whether it can be engineered—and more on whether regulators, manufacturers, and consumers decide it should be.
spectrum.ieee.org
3 months ago
This webinar looks at a Battery Electric Virtual Vehicle Model of a mid-size BEV, and uses Simulink and Simscape to facilitate design exploration, component refinement, and system-level optimization. The virtual vehicle comprises five subsystems: Electric powertrain, driveline, refrigerant cycle, coolant cycle, and passenger cabin. The model will be tested using different drive cycles, cooling, and heating scenarios. The results will be analyzed to determine the impact of the different design parameters on vehicle consumption. The resulting virtual vehicle will be used to: Test different drive cycles and environmental conditions Perform sensitivity analysis Optimize model to improve thermal performance and consumption Register now for this free webinar!
content.knowledgehub.wiley.com
3 months ago
Charles Proteus Steinmetz was a towering figure in the early decades of electrical engineering, easily the intellectual equal of Thomas Edison and Nikola Tesla—men he considered his friends. One of Steinmetz’s most significant achievements was to quantify and characterize the phenomenon of magnetic hysteresis—the behavior of magnetism in materials—and then devise a simple law that allowed for predictable transformer and motor design. He also established a revolutionary framework for analyzing AC circuits, which is still taught today in power engineering. And from 1893, he served as chief consulting engineer at General Electric at a pivotal moment for the young company and for the U.S. effort to expand its power grid. For these and other accomplishments, he was well known in his time, even if he’s not exactly a household name today. Steinmetz was also an evangelist for electric vehicles. In March 1920, he typed out his thoughts, comparing the pros and cons of EVs to the gasoline-propelled alternative. Among the advantages: low cost of maintenance, reliability, simplicity of operation, and lower cost of operation. The disadvantages: dependence on charging stations, limited range on a single charge, and lower speeds. More than a century later, his list remains remarkably pertinent. Steinmetz could often be seen decked out in a suit and top hat, smoking his trademark BlackStone panatela cigar while riding around Schenectady, N.Y., in his 1914 Detroit Electric sedan. According to John Spinelli, emeritus professor of electrical and computer engineering at Union College , in Schenectady, sometimes both Steinmetz and his chauffeur sat in the backseat—you could control the car from both the front and the rear—so that it would appear to be a driverless car. With a top speed of 40 kilometers per hour (25 miles per hour), the car ran on 14 six-volt batteries and could go about 48 km between charges. Steinmetz’s 1914 Detroit Electric car is now at Union College in Schenectady, N.Y., where Steinmetz had founded, chaired, and taught in the department of electrical engineering. Paul Buckowski/Union College In 1971, the car was purchased by Union College, where Steinmetz had founded, chaired, and taught in the department of electrical engineering. The car had been discovered rotting in a field, so it needed some work. Over the next decade, faculty and engineering students restored it to its former glory. Still in running condition, it’s now on permanent display at the college. Steinmetz’s Contributions to Electrical Engineering Karl August Rudolf Steinmetz was born in 1865 in Breslau, Prussia (now known as Wrocław, Poland). He studied mathematics, physics, and the burgeoning field of electricity at the University of Breslau. He also joined a student socialist club and edited the party newspaper, The People’s Voice . He completed his doctoral studies, but before receiving his degree, Steinmetz fled to Switzerland in 1888, after his socialist writings came under the scrutiny of the Bismarck government. Steinmetz immigrated to New York the following year, anglicized his first name, dropped his two middle names, and added Proteus, a nickname he had picked up at university (after the shape-shifting sea god of Greek mythology). Eventually, he became a U.S. citizen. Charles Proteus Steinmetz solved a number of important problems that helped the power grid expand. Bettmann/Getty Images In January 1892, Steinmetz burst onto the engineering scene when he read his paper “ On the Law of Hysteresis ” before the American Institute of Electrical Engineers, a forerunner of today’s IEEE. I can’t quite imagine sitting through the delivery of its 62 pages, but those assembled recognized its groundbreaking nature. The ideas Steinmetz outlined allowed engineers to calculate power losses in the magnetic components of electrical machinery during the design phase. Prior to this, the design process for transformers and electric motors was largely trial and error, and power losses could be measured only after the machine was built, which greatly added to the cost. Steinmetz was not just an equations and theory guy, though. He loved working in the lab and building things. In 1893, General Electric acquired the small manufacturing firm of Eickemeyer & Osterheld, in Yonkers, N.Y., where Steinmetz had worked since shortly after his arrival in the United States. So Steinmetz began his new life as a corporate engineer, an interesting turn for the socialist. During his first few years with GE, he mostly designed generators and transformers. But he also created an informal position for himself as a consultant, giving expert opinions on various problems across divisions. He eventually formalized this role, becoming GE’s chief consulting engineer, and he maintained a relationship with the company for the rest of his life, even after joining the faculty of Union College in 1902. By the time Steinmetz died in 1923 at the age of 58, he had been granted more than 200 patents and had made major contributions to various subfields in electrical engineering, including phasors and complex numbers (for steady-state AC analysis); electrical transients, switching surges, and surge protection (based on his research on lightning); industrial research (including how to run a corporate lab); and engineering methods (by writing textbooks that standardized practice). Why Steinmetz Believed in Electric Cars By 1914, Steinmetz was convinced that the future of transportation was electric. In June, he addressed the National Electric Light Association convention in Philadelphia with a bold prediction: “ I have no doubt that in 10 years, more or less—rather less than more—we will see the field of the pleasure and business vehicle covered by such an electric car in large numbers. And I believe I underestimate when I say that 1,000,000 or more will be used.” As we now know, Steinmetz was overly optimistic. At the time, there were about 1.2 million gasoline-powered cars in use in the United States, and only about 35,000 EVs. It would take until 2018 for the number of EVs (including plug-in hybrids) on U.S. roads to surpass a million. Worldwide, there are now about 60 million electric vehicles in use. But Steinmetz had his reasons. He firmly believed that electric vehicles would flourish in urban areas, where most rides involved short distances at low speed. He also thought EVs would be a boon for power companies, which were eager to drum up more business, especially at night. With 1 million electric cars being charged about 5 kilowatt-hours on most nights, and at a rate of 5 cents per kilowatt-hour, Steinmetz predicted US $75 million (about $2.5 billion today) of new business for central power stations each year. In 1971, Union College purchased Steinmetz’s car, which had been found rotting in a field, and faculty and students restored it to working condition. Special Collections & Archives/Schaffer Library/Union College Steinmetz went to work to improve the electric car. He developed a double-rotor motor that was integrated into the rear axle, which did away with the need for a mechanical differential or drive shaft and drastically reduced the overall weight, which improved the mileage. Dey Electric Corp. incorporated Steinmetz’s design into its electric roadster and priced it under $1,000. Unfortunately, an internal combustion engine Ford Model T cost about half as much, and the Dey roadster flopped, ending production within a year. Undeterred, Steinmetz formed the Steinmetz Electric Motor Car Corp. in 1920 with the initial goal of bringing to market an electric truck for deliveries and light industrial use. The first truck debuted on a cold February day in 1922 with a publicity stunt of climbing the steep Miller Avenue hill in Brooklyn, N.Y. According to a report in The New York Times, the vehicle went up the 14.5 percent grade between Jamaica Avenue and Highland Boulevard in 51 seconds. During a second climb, it stopped a number of times to show how easily it restarted. The truck had a range of 84 km (52 miles). The company planned to manufacture 1,000 trucks per year and 300 lightweight delivery cars, plus a five-passenger coupe, but it made a total of only 48 vehicles. After Steinmetz died in 1923, the company soon ceased operation. Steinmetz wasn’t only bullish on the electric car, but on electricity in general. A New York Times article recorded his belief that by 2023, we would work no more than 4 hours a day, 200 days a year because electricity would have eliminated the drudgery and unpleasantness of labor. He also predicted that electricity would bring about an end to urban pollution: “Every city would be a spotless town.” With an expansion of leisure time, people would be healthier, engaging in gardening (especially growing their own food) and pursuing educational interests to become “much more intelligent and self-expressive creature[s].” Steinmetz’s Chosen Family I decided to write about Steinmetz last year, after IEEE Spectrum published an essay I wrote about why engineering needs the humanities . The article contains this line: “In 1909, none other than Charles Proteus Steinmetz advocated for including the classics in engineering education.” I had been impressed to learn of Steinmetz’s recognition of the value of a liberal arts education. But my copy editor didn’t know who Steinmetz was or why he merited the qualifier “none other.” More people should know about this remarkable man, I decided. And so I went looking for a museum object associated with him, so I could include him in a Past Forward column. Steinmetz [left] was easily the intellectual equal of Thomas Edison [right], whom he considered a friend. Corbis/Getty Images The electric car is only one avenue into Steinmetz’s life. I could instead have looked into Steinmetz solids (the geometric shapes that form when two or three identical cylinders intersect at right angles), Steinmetz curves (the edges of a Steinmetz solid), or the Steinmetz equivalent circuit (a mathematical model that describes a transformer using resistors and inductors). But none of those concepts could be easily captured in a picture-worthy object. His love of his electric car, on the other hand, was a fun and fitting entry point for this most unusual engineer. I also saw an opportunity to highlight how Steinmetz became a family man. Steinmetz had dwarfism—he stood just 122 centimeters tall—as well as kyphosis , a severe curvature of the spine, as did his father and grandfather. He didn’t wish to pass along those traits, and so he never married or had children of his own. But that didn’t mean he didn’t want a family. In 1903, Steinmetz’s favorite lab assistant, Joseph LeRoy Hayden, told his boss that he was getting married. Steinmetz invited the couple to dinner, and then invited them to live in his large home. They agreed to this unusual living arrangement, with Corinne Rost Hayden running the household and cooking for her husband and Steinmetz. She forced the men to set aside their work for regular family meals. Eventually, the Hayden family expanded, welcoming Joe, Midge, and Billy. Steinmetz legally adopted the elder Hayden, thereby gaining three grandchildren as well. Steinmetz, whom The New York Times had named a “modern Jove” who “hurls thunderbolts at will” (from a high-voltage lightning generator), delighted at entertaining the grandkids with wondrous tricks of electricity and chemistry. In writing about the history of electrical engineering, I sometimes fall into the trap of focusing too much on the technology. But it’s just as important to recognize the people behind the technology—their personalities, their frailties, their feelings, their challenges. Steinmetz faced adversity for his political beliefs, for being an immigrant, and for his physical stature, yet none of that ever stopped him. In word and deed, he showed that he had a generous heart as mighty as his intellect. Part of a continuing series looking at historical artifacts that embrace the boundless potential of technology. An abridged version of this article appears in the March 2026 print issue as “Charles Proteus Steinmetz Loved His Electric Car.” References IEEE Power & Energy Magazine published Steinmetz’s pro/con list comparing electric cars to those with internal combustion engines in the September/October 2005 issue, along with a good biographical overview of Steinmetz by Carl Sulzberger. Union College published a nice story about the restoration of Steinmetz’s electric car in 2014, when it received its permanent home on campus. There are many biographies of Steinmetz, one published as early as 1924 , but I am particularly fond of Steinmetz: Engineer and Socialist by Ronald Kline (Johns Hopkins University Press, 1992). Gilbert King’s 2011 article “ Charles Proteus Steinmetz, the Wizard of Schenectady ” for Smithsonian magazine describes Steinmetz’s chosen family and includes several fun anecdotes not mentioned above.
spectrum.ieee.org
3 months ago
In the race to bring solid-state batteries to cars, a dark horse is breaking out of the pack: Karma Automotive plans to integrate Factorial Energy’s cells into its new Kaveya coupe in late 2027. Factorial made waves in 2025 by supplying batteries for a lightly modified Mercedes EQS production sedan that covered an impressive 1,205 kilometers (749 miles) on a single-charge drive from Stuttgart to Malmö, Sweden. Factorial, based in Massachusetts, was cofounded by Siyu Huang and her husband Alex Yu and continues to supply cells to Mercedes , Stellantis and other customers for testing and validation. But the Chinese-owned Karma, formed from the ashes of two bankrupt companies— Fisker Automotive and battery maker A123 Systems—could become the first solid-state four-wheel EV in U.S. or European showrooms if it can drag its long-gestating Kaveya over the finish line. Marques McCammon , president of Karma, based in Irvine, Calif., says Factorial’s high-profile battery would give this underdog automaker a technical and market edge in its effort to compete with better-known ultraluxury brands such as Rolls-Royce. McCammon says the company reworked its original Kaveya design after realizing that conventional high-nickel-cathode lithium-ion batteries weren’t going to reach its lofty performance targets. “It was not a formula that we were satisfied with,” McCammon says. “We were going to have to carry a lot more mass, including in terms of thermal management. We were adding either a bunch of cost or a bunch of weight that unbalanced the car.” Solid-state’s potential has dozens of companies vying for first-mover advantage. The batteries replace liquid electrolytes with a solid ceramic or glass material, with huge potential to carry more energy, charge faster, and improve fire safety in comparison with today’s lithium-ion cells. Factorial’s Battery Tech Might Reach 500 Wh/kg Factorial’s Electrolyte System Technology (FEST) is made of pouch cells that are semi-solid state, meaning that they have a largely but not entirely polymer electrolyte. Huang, Factorial’s CEO, says these cells still use a bit of electrolyte fluid to make them more compatible with current battery manufacturing processes. The cells can use a conventional graphite anode, or lithium-metal or silicon anodes for better performance. The still-evolving FEST cells have an energy density around 391 watt-hours per kilogram, as demonstrated in the Mercedes EQS. For comparison, Tesla’s commonly used 4680 battery reportedly has energy density in the range of 272 to 296 Wh/kg . Factorial is also developing a fully solid-state “Solstice” cell, with a sulfide electrolyte, which the company says could reach 500 Wh/kg, as measured at the cell level. That energy density is about 70 percent higher than today’s best high-nickel batteries, and nearly 2.5 times that of leading lithium-ion phosphate cells. Since its early “production hell” in setting up a pilot battery line near Seoul in 2022, Huang says the company has boosted FEST yields from 10 percent to 85 percent—a critical factor in reducing scrap, controlling costs, and proving commercial viability. “For a pilot line like that, even for the traditional industry, 70 to 80 percent yield is pretty good,” Huang says. “We were able to achieve 85 percent, which is state of the art for established cell makers, not to mention solid state.” The company, which has a second pilot line in Massachusetts, is aiming for annual manufacturing capacity at the Korean line of about 500,000 cells, or 200 megawatt-hours. The Kaveya coupe is the kind of pricey, limited-production car that, for now, can take best advantage of Factorial’s tech. The roughly US $400,000 Kaveya features an aluminum structure, carbon-fiber body, and fanciful butterfly doors. A dog bone–shaped battery configuration, as opposed to the conventional skateboard design, allows the Kaveya to balance its weight and sit lower to the ground than typical EVs, improving its silhouette and performance. The all-wheel-drive Kaveya is targeting over 400 km (250 miles) of range, 735 kilowatts (1,000 horsepower) of electric power, a zero to 100-kilometer-per-hour (0 to 62 mph) time of roughly two seconds, and a top speed above 322 kph (200 mph). It’s a modern software-defined vehicle with a zonal architecture, which uses powerful centralized computing to replace dozens or hundreds of individual or haphazardly located control units. Siyu Huang, CEO of Factorial, sits behind the wheel of a Karma Kaveya. At right is Marques McCammon, president and chief executive of Karma Automotive. Karma Automotive Karma plans to produce about 300 Kaveyas in its opening year, and then, if all goes well, expand its lineup and reach annual capacity of 3,000 to 5,000 cars. McCammon and Huang argue that solid-state tech will drive EVs farther into the mainstream, allowing them to shrink their battery packs, weights, and eventually costs. Alternatively, cars could maintain current-size packs with jumps in driving range. A dvantages appear especially clear for the SUVs and pickups that dominate the U.S. market but whose EV versions remain bulky and cost-prohibitive due to overscaled batteries . Ford canceled its F-150 Lightning pickup after initially strong sales stalled. Ram axed its electric Ram 1500 REV pickup, whose 229-kilowatt-hour battery would have been the global industry’s largest, before it reached showrooms. Up Next: Electrifying Muscle Cars “We cannot continue to have this excessive mass in vehicles and still get to the value propositions we want,” McCammon says, “because that is what will stall the growth of electrified power trains.” Potentially being first to market “is absolutely a coup for us,” he adds. “But our business model is to catalyze change and growth for the industry.” Factorial has also partnered with Stellantis’s Dodge brand, known for its gasoline-powered muscle cars, for a demonstration fleet of electric Charger Daytonas powered by a version of the FEST batteries. Stellantis validated these 77-ampere-hour cells, which have a claimed robust energy density of 375 Wh/kg. The cells can charge from 15 to 90 percent capacity in a swift 18 minutes, and have demonstrated more than 600 charging cycles. Huang calls solid-state technology a perfect opportunity for domestic battery makers and automakers to leapfrog current technology. She notes that lithium-ion batteries were invented in the United States, which dropped the ball and allowed Japan’s Sony to commercialize them , and Korea and China to eventually dominate their manufacture and supply chains. “It’s now a $300 billion industry, right?” she says. “So for the next 30 years, I do think we need to focus on solid state. We’ve seen enough cases in the industry that being a copycat in lithium-ion batteries doesn’t work.” Some automakers and battery leaders, including Kurt Kelty , the former Tesla battery guru who now heads General Motors’ battery technology teams, believe solid-state batteries remain several years from production. Huang acknowledges skepticism over solid-state batteries, driven by companies around the world that have trumpeted production breakthroughs that fail to materialize. But Huang says Factorial has kept its promises and met its timelines so far. Where Mercedes and Factorial targeted a 1,000-km (621-mile) range for its EQS demonstration, the Mercedes actually traveled 1,205 km (749 miles) on a charge, with 137 km of range to spare. “So the car actually overdelivered,” Huang says. “We wanted to hold that promise, and we do think we’re closer than ever” to production. “By supporting this Karma collaboration, we’re targeting to launch our first vehicle on the road in 2027 And that’s a timeline we’re holding firm on.”
spectrum.ieee.org
3 months ago
At CES 2026 in Las Vegas, Singapore-based startup Strutt introduced the EV 1 , a powered personal mobility device that uses lidar , cameras , and onboard computing for collision avoidance. Unlike manually steered powered wheelchairs , the EV 1 assists with navigation in both indoor and outdoor environments—stopping or rerouting itself before a collision can occur. Strutt describes its approach as “shared control,” in which the user sets direction and speed, while the device intervenes to avoid unsafe motion. “The problem isn’t always disability,” says Strutt cofounder and CEO Tony Hong . “Sometimes people are just tired. They have limited energy, and mobility shouldn’t consume it.” Building a mobility platform was not Hong’s original ambition. Trained in optics and sensor systems, he previously worked in aerospace and robotics. From 2016 to 2019, he led the development of lidar systems for drones at Shenzhen, China–based DJI , a leading manufacturer of consumer and professional drones. Hong then left DJI for a position as an assistant professor at Southern University of Science and Technology in Shenzhen—a school known for its research in robotics, human augmentation, sensors, and rehabilitation engineering. However, he says, demographic trends around him proved hard to ignore. Populations in Asia, Europe, and North America are aging rapidly. More people are living longer, with limited stamina, slower reaction times, or balance challenges. So, Hong says he left academia to develop technology that would help people facing mobility limitations. Not a Wheelchair—an EV EV 1 combines 2 lidar units, 2 cameras, 10 time-of-flight depth sensors, and 6 ultrasonic sensors. Sensor data feeds into onboard computing that performs object detection and path planning. “We need accuracy at a few centimeters,” Hong says. “Otherwise, you’re hitting door frames.” With the touchscreen interface, users can select a destination within the mapped environment. The onboard system calculates a safe route and guides the vehicle at a reduced speed of about 3 miles per hour. The rider can override the route instantly with joystick input. The system even supports voice commands, allowing the user to direct the EV 1 to waypoints saved in its memory. The user can say, for example, “Go to the fridge,” and it will chart a course to the refrigerator and go there, avoiding obstacles along the way. The Strutt EV 1 puts both joystick controls and a lidar view of the environment in front of the device’s user. Strutt Driving EV 1 in manual mode, the rider retains full control, with vibration feedback warning of nearby obstacles. In “copilot” mode, the vehicle prevents direct collisions by stopping before impact. In “copilot plus,” it can steer around obstacles while continuing in the intended direction of travel. “We don’t call it autonomous driving,” Hong says. “The user is always responsible and can take control instantly.” Hong says Strutt has also kept its users’ digital privacy in mind. All perception, planning, and control computations, he says, occur onboard the device. Sensor data is not transmitted unless the user chooses to upload logs for diagnostics. Camera and microphone activity is visibly indicated, and wireless communications are encrypted. Navigation and obstacle avoidance function without cloud connectivity. “We don’t think of this as a wheelchair,” Hong says. “We think of it as an everyday vehicle.” Strutt promotes EV 1 ’s use for both outdoor and indoor environments—offering high-precision sensing capabilities to navigate confined spaces. Strutt To ensure that the EV 1 could withstand years of shuttling a user back and forth inside their home and around their neighborhood, the Strutt team subjected the mobility vehicle to 2 million roller cycles—mechanical simulation testing that allows engineers to estimate how well the motors, bearings, suspension, and frame will hold up over time. The EV 1 ’s 600-watt-hour lithium iron phosphate battery provides 32 kilometers of range—enough for a full day of errands, indoor navigation, and neighborhood travel. A smaller 300-Wh version, designed to comply with airline lithium-battery limits, delivers 16 km. Charging from zero to 80 percent takes 2 hours. Might These EVs Be Covered by Insurance? The EV 1 retails for US $7,500—a price that could place it outside the reach of people without deep pockets. For now, advanced sensors and embedded computing keep manufacturing cost high, while insurance-reimbursement coverage for AI-assisted mobility devices in the United States depends on where a person lives. “A retail price of $7,500 raises serious equity concerns,” says Erick Rocha, communications and development coordinator at the Los Angeles–based advocacy organization Disability Voices United . “Many mobility device users in the United States rely on Medicaid ,” the government insurance program for people with limited incomes. “Access must not be restricted to those who can afford to pay out of pocket.” Medicaid coverage for high-tech mobility devices varies widely by state, and some states have rules that create significant barriers to approval (especially for nonstandard or more specialized equipment). Even in states that do cover mobility devices, similar types of hurdles often show up. Almost all states require prior approval for powered mobility devices, and the process can be time-consuming and documentation-heavy. Many states rigidly define what “medically necessary” means. They may require a detailed prescription describing the features of the mobility device and why the patient’s needs cannot be met with a simpler mobility aid such as a walker, cane, or standard manual wheelchair. Some states’ processes include a comprehensive in-person exam, documenting how the impairment described by the clinician limits activities of daily living such as toileting, dressing, bathing, or eating. Even if a person overcomes those hurdles, a state Medicaid program could deny coverage if a device doesn’t fit neatly into existing Healthcare Common Procedure Coding System billing codes “Sensor-assisted systems can improve safety,” Rocha says. “But the question is whether a device truly meets the lived, day-to-day realities of people with limited mobility.” Hong says that Strutt, founded in 2023, is betting that falling sensor prices and advances in embedded processing now make commercial deployment of the EV 1 feasible.
spectrum.ieee.org
3 months ago
MicroVision , a solid-state sensor technology company located in Redmond, Wash., says it has designed a solid-state automotive lidar sensor intended to reach production pricing below US $200. That’s less than half of typical prices now, and it’s not even the full extent of the company’s ambition. The company says its longer-term goal is $100 per unit. MicroVision’s claim, which, if realized, would place lidar within reach of advanced driver-assistance systems (ADAS) rather than limiting it to high-end autonomous vehicle programs. Lidar’s limited market penetration comes down to one issue: cost. Comparable mechanical lidars from multiple suppliers now sell in the $10,000 to $20,000 range. That price—roughly a tenfold drop, from about $80,000—helps explain why suppliers now are now hopeful that another steep price reduction is on the horizon. For solid-state devices, “it is feasible to bring the cost down even more when manufacturing at high volume,” says Hayder Radha , a professor of electrical and computer engineering at Michigan State University and director of the school’s Connected & Autonomous Networked Vehicles for Active Safety program. With demand expanding beyond fully autonomous vehicles into driver-assistance applications, “one order or even two orders of magnitude reduction in cost are feasible.” “We are focused on delivering automotive-grade lidar that can actually be deployed at scale,” says MicroVision CEO Glen DeVos . “That means designing for cost, manufacturability, and integration from the start—not treating price as an afterthought.” MicroVision’s Lidar System Tesla CEO Elon Musk famously dismissed lidar in 2019 as “ a fool’s errand ,” arguing that cameras and radar alone were sufficient for automated driving. A credible path to sub-$200 pricing would fundamentally alter the calculus of autonomous-car design by lowering the cost of adding precise, three-dimensional sensing to mainstream vehicles. The shift reflects a broader industry trend toward solid-state lidar designs optimized for low-cost, high-volume manufacturing rather than maximum range or resolution. Before those economics can be evaluated, however, it’s important to understand what MicroVision is proposing to build. The company’s Movia S is a solid-state lidar. Mounted at the corners of a vehicle, the sensor sends out 905-nanometer-wavelength laser pulses and measures how long it takes for light reflected from the surfaces of nearby objects to return. The arrangement of the beam emitters and receivers provides a fixed field of view designed for 180-degree horizontal coverage rather than the full 360-degree scanning typical of traditional mechanical units. The company says the unit can detect objects at distances of up to roughly 200 meters under favorable weather conditions—compared with the roughly 300-meter radius scanned by mechanical systems—and supports frame rates suitable for real-time perception in driver-assistance systems. Earlier mechanical lidars used spinning components to steer their beams, but the Movia S is a phased-array system. It controls the amplitude and phase of the signals across an array of antenna elements to steer the beam. The unit is designed to meet automotive requirements for vibration tolerance, temperature range, and environmental sealing. MicroVision’s pricing targets might sound aggressive, but they are not without precedent. The lidar industry has already experienced one major cost reset over the past decade. “Automakers are not buying a single sensor in isolation....They are designing a perception system, and cost only matters if the system as a whole is viable.” — Glen DeVos, MicroVision Around 2016 and 2017, mechanical lidar systems used in early autonomous-driving research often sold for close to $100,000. Those units relied on spinning assemblies to sweep laser beams across a full 360 degrees, which made them expensive to build and difficult to ruggedize for consumer vehicles. “Back then, a 64-beam Velodyne lidar cost around $80,000,” says Radha. Solid-State Lidar Design Challenges Lower cost, however, does not come for free. The same design choices that enable solid-state lidar to scale also introduce new constraints. “Unlike mechanical lidars, which provide full 360-degree coverage, solid-state lidars tend to have a much smaller field of view,” Radha says. Many cover 180 degrees or less. That limitation shifts the burden from the sensor to the system. Automakers will need to deploy three or four solid-state lidars around a vehicle to achieve full coverage. Even so, Radha notes, the total cost can still undercut that of a single mechanical unit. What changes is integration. Multiple sensors must be aligned, calibrated, and synchronized so their data can be fused accurately. The engineering is manageable, but it adds complexity that price targets alone do not capture. DeVos says MicroVision’s design choices reflect that reality. “Automakers are not buying a single sensor in isolation,” he says. “They are designing a perception system, and cost only matters if the system as a whole is viable.” Those system-level trade-offs help explain where low-cost lidar is most likely to appear first. Most advanced driver-assistance systems today rely on cameras and radar, which are significantly cheaper than lidar. Cameras provide dense visual information, while radar offers reliable range and velocity data, particularly in poor weather. Radha estimates that lidar remains roughly an order of magnitude more expensive than automotive radar. But at prices in the $100 to $200 range, that gap narrows enough to change design decisions. “At that point, lidar becomes appealing because of its superior capability in precise 3D detection and tracking,” Radha says. Rather than replacing existing sensors, lower-cost lidar would likely augment them, adding redundancy and improving performance in complex environments that are challenging for electronic perception systems. That incremental improvement aligns more closely with how ADAS features are deployed today than with the leap to full vehicle autonomy. MicroVision is not alone in pursuing solid-state lidar, and several suppliers—including Chinese firms Hesai and RoboSense and other major suppliers such as Luminar and Velodyne—have announced long-term cost targets below $500. What distinguishes current claims is the explicit focus on sub-$200 pricing tied to production volume rather than future prototypes or limited pilot runs. Some competitors continue to prioritize long-range performance for autonomous vehicles, which pushes the cost upward. Others have avoided aggressive pricing claims until they secure firm production commitments from automakers. That caution reflects a structural challenge: Reaching consumer-level pricing requires large, predictable demand. Without it, few suppliers can justify the manufacturing investments needed to achieve true economies of scale. Evaluating Lidar Performance Metrics Even if low-cost lidar becomes manufacturable, another question remains: How should its performance be judged? From a systems-engineering perspective, Radha says cost milestones often overshadow safety metrics. “The key objective of ADAS and autonomous systems is improving safety,” he says. Yet there is no universally adopted metric that directly expresses safety gains from a given sensor configuration. Researchers instead rely on perception benchmarks such as mean average precision , or mAP, which measures how accurately a system detects and tracks objects in its environment. Including such metrics alongside cost targets, says Radha, would clarify what performance is preserved or sacrificed as prices fall. IEEE Spectrum has covered lidar extensively, often focusing on technical advances in scanning, range, and resolution. What distinguishes the current moment is the renewed focus on economics rather than raw capability. If solid-state lidar can reliably reach sub-$200 pricing, it will not invalidate Elon Musk’s skepticism—but it will weaken one of its strongest foundations. When cost stops being the dominant objection, automakers will have to decide whether leaving lidar out is a technical judgment or a strategic one. That decision, more than any single price claim, may determine whether lidar finally becomes a routine component of vehicle safety systems.
spectrum.ieee.org
3 months ago
A superpowered Formula 1 car, a buzzing drone, a soldier’s pack, and a wearable smart device have this in common: They all need batteries. Ideally, those batteries could fit into oddly shaped nooks, curves, and voids, something that today’s cylindrical or rectangular cells struggle to do. Engineer Gabe Elias, who helped design the Mercedes-AMG Petronas racers that won seven consecutive F1 championships, cofounded a startup to 3D print batteries onto surfaces, flowing into those unused spaces in all kinds of devices and vehicles. The company recently won a US $1.25 million, 18-month contract with the U.S. Air Force to prove its tech’s potential. It joins competitors such as Sakuú , in Silicon Valley, and Germany’s Blackstone Technology, in a race to popularize printed batteries that can conform to various shapes. Soon after Elias cofounded Material Hybrid Manufacturing in 2023, his group realized that their initial pitch—printing batteries in new shapes for passenger cars —was stuck in neutral. EVs, especially bigger ones, don’t have pressing space constraints for batteries. Electric SUVs and pickup trucks from Rivian , where Elias also worked, can fit 7,776 cylindrical batteries into a brawny, 135 kilowatt-hour pack. So, the company changed lanes to smaller devices with wasted space it could stuff with energy. The Hybrid3D, a proprietary manufacturing platform, can print full-stack batteries in situ: Anode, cathode, separator and casing, with no molds or costly tooling required. The tech eliminates the metal casings, bus bars and other components that hog space in conventional cells. Material’s active battery material can fill voids and follow three-dimensional curves: Think the wing of a drone, or the slender, curling arm of a pair of smart glasses. “Things are shrinking, so we’re shrinking around it,” Elias says. “Electronics are becoming embedded, consolidated, optimized, and batteries are the only part of that equation that’s being left behind.” The company has teamed with Performance Drone Works (PDW) to push its tech toward commercialization. For the initial project, the companies will show how much active battery their 3D material can pack into the same modular space that holds 48 cylindrical cells in an existing drone. Even in that simplified, proof-of-concept drone, the printed battery achieves a 50 percent boost in energy density, and uses 35 percent more available volume. “That gives you a bunch of options,” Elias says. “You could either fly 50 percent farther, or decrease the size of the battery pack, fit more payload, and cover the same distance.” Fit for purpose Next-gen designs could boost those gains by dispersing battery material around drone frames, electric motors, or other surfaces. Notoriously heavy military backpacks—stuffed with bulky, square batteries—could be lighter and ergonomically shaped. Military helmets could directly integrate batteries that power head-mounted gear. When he was still at Mercedes, Elias tried to wrap conventional cells around a driver’s seat to improve the layout. Even in the rarefied realm of F1 racing, where top teams spend hundreds of millions of dollars a year in the service of speed, Mercedes simply gave up. “We ended up stopping the project, just knocking our heads against the wall because it’s so complicated to take these little cylindrical cells, wedge them into spaces and tie them together in the configuration you want,” he says. Material’s batteries, he says, are the natural evolution of carbon fiber or other composite structures in automobiles, including “cell to pack” construction that eliminates modules and makes batteries integral parts of structures. “We’re turning energy storage into a subsystem, just like all the other subsystems on a car,” Elias says. Material’s first commercial-scale printer’s bed measures 550 by 350 millimeters, with plans to greatly expand that surface. Its tech is essentially a hybrid of direct ink printing and fused deposition modeling , two of several techniques being developed by companies vying to bring these energy sources to market. Critically, the tech would allow batteries to go from prototype to printing, with no need for expensive, time-consuming retooling. Material’s printer platform can already handle a variety of chemistries and formats with simple changes in materials and software coding, he says. “We’ve printed NMC 811 and NMC 111, LFP and lithium-titanate oxide (LTO), to name a few,” Elias says. “We’re chemistry-agnostic.” Material’s cells currently use liquid electrolyte, added via an infusion process, but the company has a working road map toward solid-state designs. Challenges include tuning battery materials to flow properly from printer nozzles; and to deposit that material in uniform, repeatable layers, roughly 100 to 150 micrometers thick, to ensure high quality and yields. “Batteries really live and die by layer thicknesses,” Elias says. Elias notes that Apple and other companies are investing massive amounts of money to create conformable batteries , such as the L-shaped batteries in some iPhones, but are using costly and limited traditional methods. And with consumer electronics giants fighting to popularize wearable devices, printed batteries are an enticing solution for their packaging and power needs. Elias points to Apple’s Tim Cook, who has gone from an AR skeptic to a champion of smart glasses. But consumers won’t really bite, he believes, until the form factor says “Ray-Ban stylish” rather than “four-eyed dork.” “I want the connectivity and usability, but I don’t want to look stupid, or I’m just going to pull out my smartphone,” he says. “We see this as a proliferated application where everyone and their mother is going to have one of these devices.” If companies can manage to print batteries the way office workers print a document, the technology would replace much of the costly tooling, dedicated factory production lines and time-consuming processes of conventional batteries. Those printed batteries could then compete on cost across the entire market, Elias says, from single cells to complex multicell packs, where prices can range from $400 to $3,000 per kilowatt-hour. “The more complex the pack, the more value we capture from part consolidation and system integration, so those applications actually carry higher margins for us,” he says. This article appears in the April 2026 print issue as “3D Printed Batteries Fit Every Nook and Cranny.”
spectrum.ieee.org
4 months ago
More than 97 percent of the new cars Norwegians registered in November 2025 were electric, almost reaching the country’s goal of 100 percent. As a result, the government has begun removing some of the many carrots it used to encourage its successful EV transition . Cecilie Knibe Kroglund, state secretary in the country’s Ministry of Transport, reveals some of the challenges that come with success. Cecilie Knibe Kroglund Cecilie Knibe Kroglund is the state secretary in Norway’s Ministry of Transport. What were the important early steps to promote the EV switch? Kroglund: Battery-electric vehicles have had exemptions from the 25 percent value-added tax and from the CO 2 - and weight-based registration tax that apply to combustion-engine vehicles. We used other tax incentives to encourage building charging stations on highways and in rural areas. Cities had the opportunity to exempt zero-emissions cars from toll roads. EV drivers also got reduced ferry fares, free parking, and access to bus lanes in many cities. The technology for the vehicles wasn’t that good at the start of the incentives program, but we had the taxes and incentives to make traditional passenger cars more expensive. What were the biggest barriers, and how did policymakers overcome them? Kroglund: Early on the technology was challenging. In summertime it was easy to fuel the EV, but in wintertime it’s double the use of energy. But the technology has improved a lot in the last five years. The Norwegian tax exemptions on EVs were introduced before EVs came to market and were decisive in offsetting the early disadvantages of EVs compared to conventional cars, especially regarding comfort, vehicle size, and range. The rapid expansion of charging infrastructure along major corridors has also been important to overcome range anxiety. How have private companies responded to government incentives? Kroglund: I’m personally surprised that it went so well. This was a long-term commitment from the government, and the market has responded to that. Many Norwegian companies use EVs. The market for charging infrastructure is considered commercially viable and no longer needs financial support. However, we don’t see commercial-vehicle adoption going as fast as passenger vehicles, and we had the same goal. So we will have to review the goals, and we’ll have to review the incentives. What unexpected new problems is Norway’s success creating? Kroglund: The success of the passenger-vehicle policies mean EVs are in competition with public transport in the larger cities. Driving an EV remains much cheaper than driving a conventional car even without tax exemptions, and overall car use continues to rise. National, regional, and local governments must find different tools to promote walking, bicycling, and public transport because each city and region is different. How applicable are these lessons to poorer or less well-administered countries and why? Kroglund: We are different as countries. The geographies are different, and some countries have even bigger cities than our national population. This is not a policy for L.A., but what we see in Norway is that incentives work. However, tax incentives are only applicable in systems where effective taxation is established, which may not be the case in poorer countries. Other benefits, such as lower local emissions, only apply in places with lots of traffic. The Norwegian experience shows that the economic incentives work, but it also shows that EVs work even in a country with cold weather. This article appears in the February 2026 print issue as “Cecilie Knibe Kroglund.”
spectrum.ieee.org
4 months ago
In a global race to get solid-state batteries on the road, few would bet on two tiny companies in Estonia, known for their innovative hubless, in-wheel electric motors and motorcycles. Yet these upstarts have apparently done what Tesla, BYD and other EV-and-battery titans have been unable to do. To be fair, building a relative handful of batteries for a low-volume motorcycle is a whole different ball game from, say, Toyota having to validate and stand behind thousands or millions of car batteries under warranty. Nevertheless, Verge Motorcycles and its tech spin-off, Donut Lab , are claiming a checkered flag at CES 2026 in Las Vegas: The Verge TS Pro motorcycle will begin shipping with Donut Lab’s solid-state batteries in the first quarter of this year, founders of the two companies told IEEE Spectrum . All other Verge bikes will follow with their own solid-state packs, to be built in Finland, just across the Gulf of Finland from Estonia. Short riding range and frequent, lengthy charging stops have been big bummers for electric motorcycles. Their whispery hum may be welcome while pulling into a quiet subdivision at 3 a.m. But these green machines have struggled to convert riders who crave the shriek of a 1-liter sport bike at 14,000 rpm, or the “potato-potato” rumble of a Harley V-twin. Leaving the shrieking and rumbling aside, the Verge-Donut team say their bikes, motors, and batteries overcome those challenges. The TS will integrate batteries with no lithium or liquid electrolyte to conduct ions, replaced by ceramics that trim weight and improve safety, charging performance and charging speeds. Buyers can choose between a 20.2- or a 33.3-kilowatt-hour battery pack, with a claimed energy density of 400 watt-hours per kilogram. That’s a healthy jump over the roughly 200–300 Wh /kg of typical lithium EV batteries. And with its Donut Lab motor whirring away inside it, the signature hubless rear wheel of the TS Pro, like the “light cycle” from the movie Tron , will turn heads and blow minds. Twice the Range of Electric Highway Bikes The Verge TS’s Large Battery version will roam for up to 600 kilometers (370 miles), more than double the range of a typical electric two-wheeler. With a 200-kilowatt peak rate, the Verge should charge from 20 percent to near-full in less than 10 minutes. Riders will pay a premium for all this tech: A TS Pro starts at US $29,900 in the U.S., with the Large Battery adding another $5,000. Several of the world’s leading battery experts, including General Motors’ Kurt Kelty —Tesla’s former battery chief, no less—have publicly stated that solid-state tech remains years from showrooms. The same goes for other advanced battery technologies, such as lithium-metal, which is being developed by the Volkswagen-backed startup Quantumscape . Ville Piippo , Donut Lab’s cofounder and CTO, acknowledged the skepticism that shrouds the technology like so many pesky ions. That skepticism stems largely from Donut’s lack of publicized experience in designing or manufacturing batteries. In addition, the company has so far shared no technical details of what’s inside these cells. Donut acknowledges all that, and says the proof is forthcoming, including on the motorcycles it intends to begin delivering to customers in coming months. “If the world is pouring billions and billions of dollars into solid state, why haven’t they figured this out?” Piippo asks rhetorically. “The answer is the same as it is for our motors, that we are doing things a different way, and the rest of the world has focused on the wrong thing.” The Verge Motorcycles TS Pro is immediately recognizable by its hubless rear wheel. Donut Lab Donut’s latest hubless motor weighs 21 kg, barely half the weight of the previous generation, and is about half the size. It generates 102 kilowatts (137 horsepower) and a staggering 1,000 newton meters of peak torque. That allows a 3.5-second launch to 60 mph (102 km/h)—quick for an electric motorcycle, but dawdling compared with combustion-engine sport bikes or even supercars. Dragstrip battles aside, Piippo says the electric motor’s broad, relentless power band is its secret weapon on the street. “Whenever you twist the throttle, you get immediate, huge acceleration,” he says. “The torque is basically a flat line from 0 km/h all the way to 200 km/h, which is the bike’s electronically limited top speed. Marko Lehtimäki , Verge’s cofounder and CTO, acknowledges that “All the claims [by other companies] have made it hard for us to be believable, but the reality is that solid state has arrived. It cannot be purely bullshit, or otherwise we’ll be destroying our reputation with consumers,” he argues. “It cannot be purely bullshit, or otherwise we’ll be destroying our reputation with consumers.” —Marko Lehtimäki, CTO, Verge Motorcycles Lehtimäki says the companies will be happy to offer media rides, deeper tech dives or factory tours in due course. Beyond motorcycles, they are developing several joint projects, including a defense-grade tactical buggy and drone platform in conjunction with ESOX Group . “These technologies are also enabling new types of vehicles that have not been possible before,” Lehtimäki says. “That is a kind of holy grail that people have been wanting to get to.” Up Next: A True, Affordable, Electric Roadster That includes a global effort to design truly lightweight electric sports cars, a conundrum that has stymied everyone from Porsche to Tesla. There are only a couple of two-seat electric sports cars on the market now. One is the $2.5 million Rimac Nevera, which has failed to capture the hearts or wallets of wealthy supercar buyers. The China-built MG Cyberster is another, but at nearly 2,000 kilograms (4,400 pounds), that roadster is an Ozempic candidate. As with motorcycles, the problem boils down to basic physics: Either the batteries weigh too much, compromising performance, or they’re too small to deliver reasonable range. Yet CES proved a cornucopia of tech reveals from these feisty players from the Baltics. Donut Lab’s in-wheel motors—though not its solid-state battery—are set to power a limited-production sports car from Longbow. The U.K. startup introduced its Speedster, a spiritual successor to British bantamweights from the likes of Caterham, Jaguar, and Lotus. The open-air two-seater, built on an aluminum chassis, weighs a svelte 895 kg, or a bit under 2,000 lbs, allowing a 0–100 km/h sprint (0–62 mph) in 3.5 seconds, and a 275-mile driving range. Even if that weight claim proves overly optimistic, consider that a Mazda Miata—among the lightest modern mass-market sports cars—checks in at roughly 1,100 kg (just over 2,400 lbs) in soft-top guise. Longbow insists it’s ready to kick off Speedster production later this year, though only 150 copies at first, priced at £84,995. A second Roadster model claims a roughly 1,000-kg (2,200-lb) curb weight. Beyond bikes and sports cars, a new collaboration with the Watt Electric Vehicle Co. will integrate a pair of Donut in-wheel motors on the rear axle of Watt’s lightweight aluminum EV platform, Piippo says. Called PACES, for Passenger and Commercial EV Skateboard, the modular unit aims to give EV startups a turnkey platform around which they could design a vehicle with relative ease.
spectrum.ieee.org
5 months ago
Ask the average driver what they want from a car, and it isn’t 0-to-60-mile-per-hour times or Nürburgring lap records. It’s something quiet, comfortable, reliable, and inexpensive to run. On all those fronts, the electric vehicle (EV) already offers a better experience than a gasoline car. EVs are more responsive, easier to maintain, and aligned with everyone’s idea of a sustainable future. After all, no one pictures a futuristic city cloaked in exhaust fumes. Yet mass adoption isn’t driven by enthusiasts—it’s driven by the everyday buyer. And for that buyer, EVs remain too costly. Global EV sales passed roughly 20 percent of new cars in 2024, according to the International Energy Agency , but the inflection point for true mass adoption still lies ahead. Some major Western automakers are signaling caution: GM, for example, paused production of the Cadillac Lyriq and Vistiq in December and will run only a single shift at its Spring Hill, Tenn., plant through early 2026—an acknowledgment of softer near-term U.S. demand and rising costs. Meanwhile, global BEV growth is being pulled forward by China. If demand worldwide is rising while Western manufacturers slow production, the industry may be entering a major shake-out. Automakers cannot sustain a multiyear cost disadvantage against Chinese competitors, and only a handful that close that gap will emerge as long-term winners. And closing it ultimately comes down to building far cheaper batteries. To reach true mass-market penetration, EVs must match internal-combustion cars on both range and affordability—roughly 400 miles for US $20,000 to $25,000. That’s a tall order, because batteries make up about 40 percent of an EV’s cost, and the cells themselves dominate that figure. BloombergNEF ’s most recent battery-price survey found that cell manufacturing is now the single biggest determinant of whether a vehicle can be profitably priced for the mass market. Where the cost lies About 70 percent of an EV battery cell’s cost comes from materials—the cathodes and anode active materials, separators, and current collectors—and 30 percent from manufacturing , according to data from Thunder Said Energy , an Austin, Texas–based consultancy focused on energy technologies. Progress on both fronts is vital. Chemistries such as lithium-iron-phosphate (LFP) and nickel-manganese-cobalt (NMC) are steadily improving in cost and performance, and researchers are exploring cheaper current-collector materials and boosting energy density with low-cost silicon-doped anodes. But even as materials evolve, the way we build cells has changed remarkably little. Today’s “wet-coating” process still resembles how it was done decades ago: active powders mixed with toxic solvents, spread as slurries onto metal foil, and dried in industrial ovens the length of a football field. A 50-gigawatt-hour cell factory—enough for about a million EVs per year—can require 50 megawatts of continuous power just for those ovens, according to a 2022 study in the Journal of Power Sources. That’s equivalent to the electricity demand of roughly 40,000 homes , the U.S. Energy Information Administration notes. The environmental and capital costs are enormous. Rethinking the factory floor That’s why the industry’s attention is turning toward dry electrode manufacturing. In principle, eliminating solvents from electrode coating could cut both energy use and cost, while shrinking factory footprints. But getting dry coating to work at scale has proven extremely difficult. Without liquids, it’s hard to mix and spread the fine powders evenly, maintain strong adhesion, and avoid damaging the materials through heat and friction. At Anaphite, my company (which is located in Bristol, England), we’ve spent nearly five years developing an alternative we call our Dry Coating Precursor (DCP) technology. We start with low-toxicity solvents to disperse materials uniformly, then remove the solvent mechanically before dry coating. The resulting film-forming powder behaves almost like kinetic sand: granular when loose, cohesive under pressure. During manufacturing, it transforms into a smooth, flexible electrode layer that bonds tightly to its current collector. The payoff is dramatic—an 85 percent reduction in coating-process energy use, up to 40 percent lower cell-production cost, and a 15 percent smaller factory footprint, all without compromising yield or performance. These savings compound rapidly: Percentage points shaved from cell cost can determine whether a vehicle remains niche or achieves true mass-market pricing. A member of Anaphite’s Cells and Electrodes team prepares battery cells whose electrodes are made with the company’s proprietary Dry Coating Precursor for testing. Anaphite Parallel paths toward the same goal Anaphite is not alone in this pursuit. On a recent episode of the Volts podcast, Karl Littau , CTO of San Jose, Calif.–based Sakuù , described his company’s solvent-free “laser-printing” method, which he likens to “frosting a cake—without the mess.” Instead of wet slurries and ovens, Sakuù’s Kavian platform fuses dry powders directly onto foil with heat and pressure. Their approach can print electrodes of nearly any chemistry—LFP, NMC, or even formulations yet to be invented—by simply swapping material cartridges. In pilot programs, Sakuù reports that its process cuts carbon-dioxide emissions by about 55 percent, shrinks factory size by 60 percent, and slashes utility costs by more than half. Other Players in the Dry-Electrode Race AM Batteries —AM, headquartered in Billerica, Mass., uses a powder-to-electrode roll-to-roll process that sprays dry active materials directly onto foil. Unlike Anaphite’s pre-treated film-forming powder, AMB skips liquids entirely, bonding particles with a small amount of binder and pressure. It targets continuous high-throughput manufacturing rather than Sakuù’s modular printers. The company is developing pilot [AH6] lines with cell makers in North America and Asia. LiCAP Technologies —The Sacramento, Calif.–based company’s Activated Dry Electrode process forms electrode sheets under heat and pressure. LiCAP has commissioned a 300-MWh dry-coating line in California and is partnering with European equipment suppliers to scale up. The machines themselves are modular and compact—“They could go in a garage,” Littau says—allowing manufacturers to scale production by adding units rather than constructing vast, energy-hungry facilities. While Anaphite and Sakuù take different engineering routes, the destination is the same: a low-cost, low-energy, high-throughput future for battery manufacturing. Why It Matters Dry coating unlocks other advantages as well. It enables thicker electrodes, which reduce the proportion of inactive materials and increase both gravimetric and volumetric energy density. The result: batteries that offer higher range per kilogram and per cubic centimeter. Combine that with EVs’ inherent benefits—quietness, smoothness, and low operating costs—and the case for electrification becomes irresistible. Whether through DCP, Kavion, or the next breakthrough waiting in a lab somewhere, the dry-coating revolution promises to make clean mobility truly mainstream—bringing forward the day when buying an EV isn’t just the cleaner choice; it’s the obvious one.
spectrum.ieee.org
5 months ago
When Sergey Antonovich rediscovered a childhood passion for music, he found an unexpected application for his skills as an embedded systems engineer: building bespoke digital accordions. Antonovich admits the accordion isn’t the coolest instrument. It was chosen for him by his mother when he was 8, and he quickly lost interest as a teenager. While growing up close to Moscow, his adolescent passions were instead channeled into electronics and tinkering with gadgets in after-school classes. This led to a career working on environmental-monitoring devices, sensors for commercial drones, and most recently, sensor systems at autonomous-vehicle developer Avride ’s R&D hub in Austin, Texas. Sergey Antonovich Employer Avride Occupation Embedded systems developer Education Master’s degree in engineering physics, Moscow Engineering Physics Institute But when Antonovich picked his accordion back up as an adult, he discovered both latent musical skills and a newfound appreciation for the instrument. Like any good tinkerer, he had some ideas about how he could improve it and he soon began using his electronics knowledge to build custom devices. And Antonovich says he’s found a surprising harmony between his day job and his hobby. Whether you’re ensuring that an autonomous vehicle spots obstacles in the road in time or translating a musician’s nimble finger work into a melodious tune, you need to rapidly process digital signals from the underlying hardware. “Both systems, self-driving cars and accordions, are real-time embedded systems,” says Antonovich. “A self-driving car is more complicated because it contains many more components, but the principles are more or less equal.” Electronics Trumps Music Antonovich grew up in Chekhov, a small town outside Moscow, and says he had a pretty ordinary childhood. His father passed away when he was only 1, so he was brought up by his mother, who worked in the printing industry, and his grandmother, a school principal who taught Russian. At 8 he was enrolled in a local music school where he learned the fundamentals of music theory and the accordion. He was a dutiful student, he says, but never felt much passion for the instrument his mother picked for him and stopped playing when he was about 15. Sergey Antonovich shows off the digital instruments he makes in his free time. With one lightweight instrument, he becomes a one-man band. Sergey Antonovich That was also when Antonovich had his first encounter with the world of electronics. He started attending after-school classes where he was taught to solder and build simple electronic systems. Antonovich quickly caught the bug and was soon knocking together digital doorbells, code locks, and basic radio receivers in his spare time. His family encouraged him to enroll at a technical secondary school, which taught engineering skills alongside the standard curriculum. When it came to picking a university, he decided he wanted a grounding in physics, so he enrolled at the Moscow Engineering Physics Institute in 2004, choosing a program that taught a combination of hardware, software, and digital-signal processing. Antonovich originally planned to become a software developer but quickly fell in love with hardware. “When you develop software, there is a level of abstraction between you and the thing itself,” he says. “But when you work with hardware, you understand how this particular thing actually works.” Embedding Into a Career Toward the end of his degree studies, in 2009, Antonovich started working for Moscow-based Ecosfera , a company focused on environmental and labor-safety measurement devices. He continued working for the company after graduating in 2010, designing hardware and software to measure conditions like temperature, humidity, and wind speed to ensure safe workplaces. It was a niche field, but one with strict regulatory requirements, and he had to shepherd his devices through rigorous certification procedures, the first major accomplishment of his career. From there, he worked for a variety of companies on different embedded and Internet of Things systems, including ATMs, medical devices, sensors for commercial drones, and digital price tags. Then in 2021 he interviewed at internet company Yandex , which operates Russia’s most popular search engine, to work on its autonomous vehicle program. “I remember I was approaching the office entrance and I saw a car which was driving itself,” Antonovich says. “You see it on YouTube, but it’s not such an inspiring experience. It’s really inspiring when you see it live.” He got the job and started work as a software engineer developing vehicle sensor systems and testing infrastructure. A corporate restructuring saw Yandex’s autonomous vehicle division spun off as a new company called Avride. Antonovich worked for the company in Israel for about a year, then in 2024 moved to its new headquarters in Austin. Antonovich says he works at Arvide primarily on the data that feed the vehicle’s perception algorithms, which includes radar and lidar . Both kinds of sensors have strengths and weaknesses—radar has long range but low resolution, while lidar is great at picking out shapes but only up to a certain distance—so the algorithmic perception system combines the data. Antonovich’s job is to build the diagnostic systems that ensure these sensors are working in perfect synchrony and deliver data within tight time limits. In his day job, Antonovich works on the sensor systems for self-driving cars. Sergey Antonovich Moving to the United States has been a positive change for Antonovich. On a professional front, the country’s soft-touch regulatory approach toward autonomous vehicles has allowed the company to make rapid progress on its technology. But he says the move has also helped him indulge his tinkering instincts in his spare time. “As a maker, I would say [the United States] is a paradise,” he says. “Electronic components are very accessible. You just order them and they arrive very quickly and everything just works.” Antonovich has taken full advantage of this to dive into his other passion—building musical instruments. A Musical Repris In 2017, when he was still living in Russia, Antonovich noticed a new generation of digital accordions emerging and it sparked his curiosity. “I thought, why not try to modify my own [acoustic] accordion?” he says. He dusted off his instrument and was gratified to find that he could still play and read sheet music. So, he tried to tackle some of the problems that beset digital accordions. Commercially available instruments are typically large and heavy, rely on bulky external modules to add musical accompaniment such as a drum beat, and connect to amplifiers with wires that restrict the performer’s movement. “I decided that maybe I can build a self-contained device,” he says. Starting with an acoustic accordion as a base, he added a synthesizer, installed internal microphones to capture acoustic sounds that could then be blended with digital ones, and integrated wireless transmitters that could free performers from cables and let them move about the stage freely. Surprisingly, Antonovich found a lot of overlap with his work on self-driving cars—in particular, the need to manage latency along the signal-processing chain. To provide a seamless experience to the player, a digital accordion needs to rapidly route input from dozens of buttons and keys on two separate keyboards to the synthesizer, which has its own processing delay. “Your main task as a developer is to keep latency as low as possible,” he says. “A high quality system should produce sound in less than 10 milliseconds, and if you come over this threshold it’s very uncomfortable to play.” Antonovich now has a growing menagerie of both hybrid acoustic-digital and fully digital accordions. But while he’s built accordions for friends, he’s in no hurry to turn his hobby into a business. “Making them a commercial product will turn my curiosity to necessity,” he says. “When you do something for a living, you do it because you have to and not because you choose to.”
spectrum.ieee.org
5 months ago
Charging an EV at home doesn’t seem like an inconvenience—until you find yourself dragging a cord around a garage or down a rainy driveway, then unplugging and coiling it back up every time you drive the kids to school or run an errand. For elderly or disabled drivers, those bulky cords can be a physical challenge. As it was for smartphones years ago, wireless EV charging has been the dream. But there’s a difference of nearly four orders of magnitude between the roughly 14 watt-hours of a typical smartphone battery and that of a large EV. That’s what makes the wireless charging on the 108-kilowatt-hour pack in the forthcoming Porsche Cayenne Electric so notable. To offer the first inductive charger on a production car, Porsche had to overcome both technical and practical challenges—such as how to protect a beloved housecat prowling below your car. The German automaker demonstrated the system at September’s IAA Mobility show in Munich. This article is part of our special report Top Tech 2026 . With its 800-volt architecture , the Cayenne Electric can charge at up to 400 kW at a public DC station, enough to fill its pack from 10 to 80 percent in about 16 minutes. The wireless system delivers about 11 kW for Level 2 charging at home, where Porsche says three out of four of its customers do nearly all their fill-ups. Pull the Cayenne into a garage and align it over a floor-mounted plate, and the SUV will charge from 10 to 80 percent in about 7.5 hours. No plugs, tangled cords, or dirty hands. Porsche will offer a single-phase, 48-ampere version for the United States after buyers see their first Cayennes in mid-2026, and a three-phase , 16-A system in Europe. Porsche’s Wireless Charging is Based on an Old Concept The concept of inductive charging has been around for more than a century. Two coils of copper wire are positioned near one another. A current flowing through one coil creates a magnetic field, which induces voltage in the second coil. In the Porsche system, the floor-mounted pad, 78 centimeters wide, plugs into the home’s electrical panel. Inside the pad, which weighs 50 kilograms, grid electricity (at 60 hertz in the United States, 50 Hz in most of the rest of the world) is converted to DC and then to high-frequency AC at 2,000 V.The resulting 85-kilohertz magnetic field extends from the pad to the Cayenne, where it is converted again to DC voltage. The waterproof pad can also be placed outdoors, and the company says it’s unaffected by leaves, snow, and the like. In fact, the air-cooled pad can get warm enough to melt any snow, reaching temperatures as high as 50 °C. The Cayenne’s onboard charging hardware mounts between its front electric motor and battery. The 15-kg induction unit wires directly into the battery. In most EVs, plug-in (conductive) AC charging tops out at around 95 percent efficiency. Porsche says its wireless system delivers 90 percent efficiency, despite an air gap of roughly 12 to 18 cm between the pad and vehicle. Last year, Oak Ridge National Laboratory transferred an impressive 270 kilowatts to a Porsche Taycan with 95 percent efficiency. “We’re super proud that we’re just below conductive AC in charging efficiency,” says Simon Schulze, Porsche’s product manager for charging hardware. Porsche also beats inductive phone chargers, which typically max out at about 70 percent efficiency, Schulze says. When the car gets within 7.5 meters of the charging pad, the Cayenne’s screen-based parking-assist system turns on automatically. Then comes a kind of video game that requires the driver to align a pair of green circles on-screen, one representing the car, the other the pad. It’s like a digital version of the tennis ball some people hang in their garage to gauge parking distance. There’s ample wiggle room, with tolerances of 20 cm left to right, and 15 cm fore and aft. “You can’t miss it,” according to Schulze. Induction loops detect any objects between the charging plate and the vehicle; such objects, if they’re metal, could heat up dangerously. Radar sensors detect any living things near the pad, and will halt the charging if necessary. People can walk near the car or hop aboard without affecting a charging session. Christian Holler, Porsche’s head of charging systems, says the system conforms to International Commission on Non-Ionizing Radiation Protection standards for electromagnetic radiation. The field remains below 15 microteslas, so it’s safe for people with pacemakers , Porsche insists. And the aforementioned cat wouldn’t be harmed even if it strayed into the magnetic field, though “its metal collar might get warm,” Schulze says. The Porsche system’s 90 percent efficiency is impressive but not record-setting. Last year, Oak Ridge National Laboratory (ORNL) transferred 270 kW to a Porsche Taycan with 95 percent efficiency, boosting its state of charge by 50 percent in 10 minutes. That world-record wireless rate relied on polyphase windings for coils, part of a U.S. Department of Energy project that was backed by Volkswagen, Porsche’s parent company. That effort, Holler says, spawned a Ph.D. paper from VW engineer Andrew Foote . Yet the project had different goals from the one that led to the Cayenne charging system. ORNL was focused on maximum power transfer, regardless of cost, production feasibility, or reliability, he says. By contrast, designing a system for showroom cars “requires a completely different level of quality and processes,” Holler says. High Cost Could Limit Adoption Cayenne buyers in Europe will pay around €7,000 (roughly US $8,100) for the optional charger. Porsche has yet to price it for the United States. Loren McDonald, chief executive of Chargeonomics, an EV-charging analysis firm, said wireless charging “is clearly the future,” with use cases such as driverless robotaxis, curbside charging, or at any site “where charging cables might be an annoyance or even a safety issue.” But for now, inductive charging’s costly, low-volume status will limit it to niche models and high-income adopters, McDonald says. Public adoption will be critical “so that drivers can convenience-charge throughout their driving day—which then increases the benefits of spending more money on the system.” Porsche acknowledges that issue; the system conforms to wireless standards set by the Society of Automotive Engineers so that other automakers might help popularize the technology. “We didn’t want this to be proprietary, a Porsche-only solution,” Schulze says. “We only benefit if other brands use it.”
spectrum.ieee.org
5 months ago
