Cracking the code on autonomous trucking
Get the full intelligence
Search transcripts, export clips, track mentions, and explore all topics from “Cracking the code on autonomous trucking” inside PodZeus.
In this episode of Catalyst, host Shayle Kann explores the challenges and opportunities in autonomous trucking with Eyal Cohen, founder and CEO of Humble Robotics. While passenger vehicle autonomy has advanced rapidly—exemplified by Waymo’s widespread deployment—autonomous trucking has lagged despite intuitive assumptions that highways would be easier than city streets. Cohen explains that the high speeds, massive weight (up to 80,000 pounds), and minimal margin for error on highways make edge cases far more dangerous than in urban environments. He traces the evolution of the industry from 2016-era efforts like Otto and Starsky Robotics, which underestimated the complexity of highway autonomy, to today’s AI-driven advancements. A key differentiator for Humble is starting with a clean-sheet design: building a cab-less, electric, fully autonomous Class A truck that combines tractor and trailer into a single intelligent platform. This design enables full rear visibility, automated safety features like warning triangle deployment, and better integration of multi-sensor data. Cohen emphasizes that while the technology is ready, unit economics remain a hurdle—especially with expensive sensors and the need for charging infrastructure. However, he believes electric short-haul and drayage operations are poised for growth, particularly with supportive infrastructure. The regulatory landscape is evolving, with states like Texas and California opening pathways for driverless testing, though new vehicle designs like Humble’s face unique challenges. Ultimately, the path to adoption hinges on proving cost savings and reliability in a B2B freight market where novelty won’t drive scale. Key takeaways include: 1) Highway autonomy is harder than city driving due to high speeds and low safety margins; 2) Vision Language Models (VLMs) are enabling camera-first perception, reducing reliance on expensive LiDAR; 3) A clean-sheet vehicle design—removing the cab and integrating tractor/trailer—can lower costs and improve safety; 4) Electrification and autonomy are synergistic, especially for short-haul, automated freight; 5) Regulatory progress is happening, but new vehicle forms require creative dialogue with agencies; 6) Unit economics must improve to justify higher vehicle costs; 7) Charging infrastructure is a chicken-and-egg problem, but solutions are emerging at ports and depots; 8) The future of freight lies in fully hands-off, automated systems from warehouse to destination.
Highway autonomy is harder than city driving due to high speeds, massive vehicle weight, and minimal margin for error.
Vision Language Models (VLMs) enable camera-first perception, reducing reliance on expensive LiDAR and accelerating AI development.
A clean-sheet vehicle design—removing the cab and integrating tractor and trailer—improves safety, visibility, and cost efficiency.
Electrification and autonomy are synergistic, especially for short-haul and drayage operations where automated charging is feasible.
Unit economics remain a major challenge; autonomous trucks must save enough on labor to offset higher vehicle costs.
…and 3 more takeaways available in PodZeus
The Autonomy Gap: Why Trucks Lag Behind Cars
Shayle Kann sets the stage by contrasting the rapid progress in passenger vehicle autonomy (e.g., Waymo) with the stalled development of autonomous trucking, despite the intuitive belief that highways should be easier than city streets.
The 2016 Myth: Why Highways Were Supposed to Be Easier
Eyal Cohen recounts his early days at Otto, where he believed highway autonomy would be simpler than urban driving. He explains the misconceptions that led to underestimating the complexity of edge cases on highways.
The Real Challenges of Highway Autonomy
Cohen details why highways are harder than expected: high speeds, long stopping distances, dangerous consequences of stopping, and rare but critical edge cases that are hard to train for.
The AI Revolution: From Hand-Coded Algorithms to VLMs
The episode explores how the evolution of AI—from hand-coded lane detection to modern Vision Language Models (VLMs)—has transformed autonomy development, enabling more intelligent perception with less manual engineering.
Sensor Fusion vs. Camera-Only: The Debate in 2026
Cohen argues for a multi-sensor approach (camera, LiDAR, radar) for safety, especially in heavy trucks, while acknowledging that VLMs are making camera-only systems more viable over time.
“The simplest possible vehicle to move freight is a box with wheels. That’s the long arc of history, and we’re moving it forward.”
“We’re not just building a truck. We’re building the future of freight: fully automated, hands-off, from warehouse to destination.”
“If you only make the tractor smart, you can’t see behind you. You can’t back into a dock. You can’t deploy warning triangles.”
Host
Guest
Eyal Cohen
person
Humble Robotics
organization
Waymo
organization
Otto
organization
Tesla
organization
Energy Hub
organization
Starsky Robotics
organization
Aurora
organization
China
place
Texas
other
Building a domestic nuclear fuel supply chain
Catalyst with Shayle Kann • 38m • 4/2/2026
Frontier Forum: Why clean energy capital boomed in a volatile year [partner content]
Catalyst with Shayle Kann • 33m • 4/6/2026
The rise of flexible data centers
Catalyst with Shayle Kann • 43m • 4/9/2026
How Base Power plans to use its fresh $1B [re-published]
Catalyst with Shayle Kann • 39m • 4/16/2026
Live from Transition-AI 2026: Inside Google’s massive AI CapEx
Catalyst with Shayle Kann • 35m • 4/23/2026
Get the full intelligence
Search transcripts, export clips, track mentions, and explore all topics from “Cracking the code on autonomous trucking” inside PodZeus.
Start discovering podcast insights today
Start with a 7-day trial and explore a growing catalog of popular podcasts. No credit card required.
No credit card required • 7-day trial • Cancel anytime
