While most autonomous trucking ventures have spent the better part of a decade bolting sensors and software onto existing semi-trucks, a San Francisco startup that emerged from stealth on April 21 is taking a different route—one that involves rethinking what a truck should look like when there's no human inside.
Humble Robotics announced its fully autonomous, cabless electric hauler alongside a $24 million seed round led by Eclipse, with participation from Energy Impact Partners. The bet: that retrofitting conventional trucks for autonomy is fundamentally the wrong approach, and that the future of short-haul freight requires hardware built from the ground up for environments humans were never meant to navigate alone.
"Trucks were never designed to be autonomous," founder and CEO Eyal Cohen told Fortune. It's a simple observation, maybe obvious in hindsight, but one that most of the industry has spent years trying to engineer around rather than confront head-on.
No Steering Wheel, No Windshield, No Driver
The Humble Hauler doesn't resemble a traditional semi in any meaningful way. There's no cab—just a narrow front wall mounting 360-degree sensor pods (cameras, LiDAR, radar) and an electric powertrain beneath a flatbed configured for 40-foot international containers or 53-foot domestic trailers. The company claims the platform is 20 percent lighter than a conventional tractor-trailer doing the same job, though it hasn't disclosed how that math works in practice.
Performance specs are modest by highway standards: 200-mile maximum range, 55 mph top speed, DC fast charging. But those numbers aren't calibrated for cross-country runs. Humble is targeting dock-to-dock freight—warehouses, railyards, seaports—where maneuverability matters more than endurance and where the economics of repetitive, short-distance hauls look very different from long-haul trucking.
The design includes multiple wheelbase variants and what Humble calls a "universal lock-and-twist interface," meaning the platform can theoretically be towed by any Class 8 tractor if something goes wrong. Though if everything goes according to plan, that feature should never get used.
The Vision-Language Bet
Humble's technology stack centers on vision-language-action (VLA) foundation models, a departure from the traditional computer vision and HD mapping approaches that dominate the sector. The company emphasizes "humanlike reasoning" and plain-language decision explanations—an architecture that, if it works, could sidestep one of autonomy's persistent headaches: the need for exhaustive pre-mapping of every route.
Humble asserts its system requires no HD mapping or predetermined route programming. That's a bold claim. Most autonomous vehicle companies have learned the hard way that real-world deployment is messier than controlled testing, and the ones that promised to skip mapping have often quietly walked that back.
The 360-degree sensor coverage is table stakes for Level 4 ambitions. The VLA layer is where Humble is placing its architectural bet—and where it will either differentiate itself or discover why everyone else took a different path.
The company built its first prototype in six months. That's an aggressive timeline, suggesting either significant prior work, exceptional team velocity, or both. Probably both.
Otto Redux
Cohen's resume reads like a tour through autonomous freight's first generation. He was part of Otto, the startup that completed the first autonomous freight delivery by semi-truck in 2016—a milestone that generated breathless headlines before the company was absorbed into Uber's ultimately doomed self-driving program. He co-founded SparkAI, an agricultural autonomy venture acquired by John Deere in March 2023, and most recently led hardware development at Waabi. Earlier stops include Apple and Uber.
Drew Gray, Humble's Head of Autonomy, brings his own pedigree: senior engineering roles at Tesla, Cruise, and Otto/Uber ATG, plus a stint as CTO at Voyage. The leadership roster also includes Ray Shan (Head of Vehicle Development), Val Eisen (Head of Legal), Siddharth Narang (Head of Hardware Technologies), and Bob Chappuis (Head of Commercialization).
As of June 2026, the company employed fewer than 50 people. That's lean for a hardware startup attempting to reinvent a category.
Eclipse partner Jiten Behl, who joined Humble's board, framed the investment thesis in the firm's announcement: "Humble understands that autonomous trucking isn't just a software problem—it requires a full-stack rethink." The framing positions Humble against competitors like Aurora and Kodiak, both of which are pursuing hub-to-hub long-haul strategies with modified conventional trucks.
California Opens the Door (Slightly)
Timing, as they say, is everything. One week after Humble's announcement, the California DMV adopted new regulations authorizing permits for heavy-duty autonomous vehicles—those with a gross vehicle weight rating above 10,001 pounds. The rules require safety driver testing first, followed by a 500,000-mile proving period for each stage before driverless deployment. Stricter reporting requirements and emergency geo-fencing authority are part of the package.
Cohen told the Los Angeles Times in June that the company had initially planned testing in Texas but was now refocusing on California, expressing particular interest in the ports of Los Angeles and Long Beach. As of late June, however, Humble hadn't yet applied for a California DMV autonomous vehicle permit. Cohen said pilots could begin "within the year."
The cabless design may face added scrutiny compared to retrofitted trucks that retain steering wheels and manual controls as fallbacks. It's a higher-stakes regulatory path. But that's the point—Humble's entire thesis is that half-measures won't cut it.
What's Missing
Humble has disclosed neither vehicle pricing, manufacturing partners, battery suppliers, nor the identities of the "market leaders in logistics and supply chain" it claims as pilot partners. The $24 million seed will fund next-generation vehicle development, autonomy system expansion, initial pilots, and early manufacturing. What it won't answer is how far that capital stretches or what the unit economics look like.
Axios reported that autonomous trucking adoption ultimately hinges on per-mile cost competitiveness; the average human-driven cost sits around $2.26 per mile. Humble's dock-to-dock model targets different economics than long-haul freight, but the company hasn't shown its math publicly.
The competitive landscape is getting crowded. Einride raised a $113 million PIPE in February ahead of a U.S. public debut and secured NHTSA approvals for cabless autonomous electric pods. Kodiak completed a SPAC merger in 2025 and is targeting driverless long-haul operations. Outrider is pushing yard autonomy with electric kits for existing terminal tractors. Each represents a different architectural bet on what autonomous freight will look like—and who will pay for it.
Starting From Zero
Humble is wagering that the future of short-haul, high-frequency freight doesn't look like a truck at all, at least not one a person would recognize. It's a cleaner thesis than trying to teach a 40-year-old vehicle platform new tricks, and it may be the only way to unlock the economics autonomy has always promised but rarely delivered.
Whether logistics operators agree will depend on pilots that haven't started yet, economics that haven't been disclosed, and regulations still being written. In an industry littered with ambitious promises and deferred timelines, Humble is at least asking a different question: What if we stopped trying to make old trucks autonomous and just built new ones instead?
The answer, of course, is that building from scratch is harder. But perhaps that's been the problem all along—everyone chose the easier path.
