A new wave of early-stage startups, backed by recent venture capital investments, is carving a niche in the autonomous vehicle sector. Unlike the companies focusing on city robotaxis, these startups are taking their technology off-road.
Two recent entrants, Seattle-based Overland AI and New Brunswick-based Potential, are gaining early advantage in this segment of autonomy. Although they apply their tech differently, both share the belief that they have solved one of the toughest challenges of automated driving by creating software that doesn’t rely on detailed maps, vast amounts of training data, or remote assistance.
The U.S. Department of Defense and venture capital investors are taking notice.
Overland AI is developing a self-driving system for military operations such as reconnaissance, surveillance, and delivering electronic warfare packages. In April, the startup received up to $18.6 million from the U.S. Army’s Defense Innovation Unit. The funds will be used to develop a prototype autonomous software stack for the Robotic Combat Vehicle (RCV) program over the next two years.
The startup, founded in 2022, also raised a $10 million seed round led by Point72 Ventures this week. These funds will be used to expand their team and further develop their autonomy stack, OverDrive, according to CEO and founder Byron Boots.
Meanwhile, Potential is creating advanced driver assistance systems (ADAS) for ATVs, underground mining vehicles, and passenger cars to handle off-road environments. The company has raised an additional $2 million CAD (~$1.5 million USD) in its seed round led by Brightspark Ventures, bringing its total funding to $8.5 million CAD (~$6.2 million USD). After six years of developing its technology, Potential is now conducting several pilot projects in power sports, motorcycles, and automotive sectors.
Off-road opportunity
Potential and Overland AI aren’t alone in applying autonomous vehicle technology to off-road areas. The high costs associated with commercial robotaxi and self-driving truck operations have led many startups to pivot their focus. As those ventures falter, new startups like Polymath Robotics, Forterra, Pronto.ai, Bear Robotics, and Outrider are emerging, targeting warehouses, mining, industrial, and off-road environments.
“We are absolutely deploying capital in off-road autonomy,” said Alexei Andreev, managing director at Autotech Ventures. He noted that most investments are in the agricultural and construction sectors, involving autonomous mining vehicles, forklifts, and tractors. These sectors aim to address labor shortages, boost productivity, and enhance safety.
Removing people from these operations can also lower insurance premiums. “The ROI for those vertical applications is now and it’s significant,” said Andreev.
Off-road autonomy has also found an ally in defense.
Overland AI: From DARPA to seed funding
The U.S. Army is a prime customer for automating off-road driving, having initiated autonomous vehicles as a DARPA project. DARPA (Defense Advanced Research Projects Agency) is a U.S. Department of Defense agency focused on advancing military technology.
“The hype around AV moved a lot of the industry toward bigger potential commercial applications, but DoD projects have persisted,” said Jeff Peters, a partner at Ibex Investors, noting that companies like SafeAI and Kodiak Robotics have also pursued defense grants. “I think AV companies will chase DoD projects because it offers large, non-dilutive funding ahead of commercial operations.”
Overland AI is an outcome of the DARPA program. Byron Boots, a professor at the University of Washington, has a long history of collaborating with the U.S. Army Research Lab and DARPA. Overland was spun out from Boots’ research and involvement in DARPA’s RACER (Robotic Autonomy in Complex Environments with Resiliency) program, aiming to develop self-driving vehicles for tough terrain.
The program is ongoing, and Overland, staffed by tech veterans from Google, Nvidia, Apple, Waymo, Aurora, Embark, Argo, SpaceX, RTX, and the U.S. Army, has been selected for its second phase.
“The high-level idea is that currently just about every ground vehicle that the military uses has a person inside of it,” said Boots. “You can imagine if you can just pull the person out of the vehicle, that confers safety and tactical advantages.”
To do so, vehicles must autonomously navigate complex terrain using only onboard sensors (mainly cameras) and computing, without relying on maps, GPS, or remote operators. Overland’s software must understand the ground’s geometry, including vegetation and mud, and how it affects vehicle dynamics.
“The terrain gets a vote on how the vehicle moves,” said Boots.
Overland’s technology processes sensor data to build a terrain representation in real-time. The vehicle uses this digital model, along with its goal location, to find a route through the terrain.
“Part of the benefit of having an autonomous system is that if you lose communication with that ground vehicle, it will continue towards its goal and try to complete the task until communication is reestablished,” said Boots.
On-road driving relies heavily on telecommunication links to remote assistance due to higher risks to other road users. This reliance is why companies like Waymo and Cruise have robotaxis stalled on streets awaiting a remote operator’s input.
“Military ground systems need to function in dynamic, unstructured terrain. Self-driving technology designed for well-defined streets and enclosed lots will struggle there; it takes a strong team to deliver effective ground autonomy in these environments,” said Chris Morales, partner at Point72 Ventures.
Potential’s potential with off-road ADAS
“How can you enable someone who maybe isn’t an expert driver to experience off-roading in challenging conditions?” asked Sam Poirier, CEO of Potential.
Potential’s core platform, Terrain Intelligence, uses computer vision to help vehicles interpret and prepare for complex terrain and changing surface conditions. The system works with data from a single camera, eliminating the need for additional sensors like lidar or radar.
At a basic level, Potential’s off-road ADAS alerts drivers about impassable objects or the need to switch drive settings based on terrain. “The next level is automating these driver-assisted settings,” said Poirier, referring to drive modes like two-wheel drive, four-wheel drive, sand, and mud modes. The final level involves using existing sensor data to optimize performance beyond what a driver can manually achieve.
“There are things assistance tools can do that individual drivers cannot,” said Scott Kunselman, ex-Jeep chief engineer and advisor to Potential. “For example, stability controls require independent brake control, which isn’t possible with a single brake pedal but can be achieved via automated systems to enhance stability.”
Yaw, the shift of a vehicle’s weight from its center of gravity to the sides, can cause spinning or fishtailing — a challenge Potential aims to address with its technology.
Potential is working with Tier 1 suppliers and OEMs to integrate its software directly into vehicles. Andreev advises focusing on relationships with Tier 1 suppliers, as OEMs may be less inclined to invest in a small startup.
