Venture capitalists and the military fuel self-driving startups that don’t need roads

A new generation of early-stage startups – as well as some recent venture capital investments – illustrate an emerging niche in the autonomous vehicle technology sector. Unlike companies bringing robo-taxis to city streets, these startups are taking their technology off-road.

Two new entrants – Seattle-based Overland AI and New Brunswick-based Potential – are poised to gain a first-mover advantage in this autonomy segment.

Although these startups apply their technology in different ways, Overland AI and Potential share common ground off-road. The founders of each startup believe they have cracked the code of one of the most complex automated driving applications by creating software that does not rely on some of the key testing and deployment crutches, such as detailed maps, vast expanses of training data. and the possibility of using remote assistance.

The U.S. Department of Defense and venture capitalists are taking note.

Overland AI, which is developing an autonomous driving system designed for military operations such as reconnaissance, surveillance and provision of electronic warfare systems, received $18.6 million from Defense in April US Army Innovation Unit. The funds will be used to build a prototype autonomous software stack for its Robotic Combat Vehicle (RCV) program over the next two years.

The startup, founded in 2022, raised a $10 million seed round this week led by Point72 Ventures. The funds will be used to expand the Overland team and continue to develop OverDrive, the company’s autonomy stack, according to CEO and founder Byron Boots.

Meanwhile, Potential, which makes advanced driver assistance systems (ADAS) that enable ATVs, underground mining vehicles and passenger cars to handle off-road environments, has raised a CA$2 million extension ( ~US$1.5 million) for its seed round led by Brightspark Ventures, a Canadian early-stage venture capitalist. This brings Potential’s total funding to $8.5 million CAD (~$6.2 million USD). The startup has spent the last six years developing its technology and is currently running several pilot projects in the powersports, motorcycle and automotive fields.

Off-road opportunity

Potentiel and Overland AI aren’t the only companies attempting to apply autonomous vehicle technology to areas outside of public streets. The costly pursuit of commercial robotaxi and self-driving truck operations has stymied dozens of startups in recent years. As these closed, a new batch of startups such as Polymath Robotics, Forterra, Pronto.ai, Bear Robotics and Outrider emerged with more grounded ambitions: applying AV technology to warehouses, mines, industrial environments and beyond. road.

“We are absolutely deploying capital into off-road autonomy,” Alexei Andreev, managing director of Autotech Ventures, told TechCrunch. “In fact, we are staying away from highway autonomy and have completely doubled off-road autonomy.”

Most of the off-road companies Autotech Ventures invests in today are in the agriculture and construction sectors, with products such as autonomous mining vehicles, forklifts and tractors. Andreev says that for these sectors, it’s about addressing labor shortages while increasing productivity and making farms and construction zones safer.

“And if you remove people, you immediately benefit from a reduction in your insurance premiums. The return on investment for these vertical applications is therefore immediate and it is significant,” said Andreev.

Another result: off-road autonomy has found a friend in defense.

Overland AI: from DARPA to seed funding

When it comes to automating off-road driving, the US military can be a great customer. After all, autonomous vehicles started as a DARPA project, says Jeff Peters, a partner at Ibex Investors. DARPA (Defense Advanced Research Projects Agency) is an agency of the United States Department of Defense focused on advancing technology for military use.

“The hype around AV has pushed much of the industry toward larger potential commercial applications, but DoD plans have persisted,” Peters told TechCrunch via email, noting that the autonomous mining startup SafeAI and autonomous trucking startup Kodiak Robotics have also applied for defense grants. “I think AV companies (those that still exist) will pursue DoD projects because they provide significant, non-dilutive funding in the interim before commercial operations.”

Overland AI is the latest by-product of the DARPA program. Boots, a professor of machine learning at the University of Washington and founder of the Robot Learning Laboratory in the university’s School of Computer Science and Engineering, has a long history of collaboration with the U.S. Army Research Laboratory and the DARPA.

Overland comes from research by Boots and the team involved in DARPA’s RACER (Robotic Autonomy in Complex Environments with Resiliency) program, which aims to develop autonomous vehicles capable of handling difficult terrain.

The program is still ongoing. Overland, which includes advanced technology veterans from Google, Nvidia, Apple, Waymo, Aurora, Embark and Argo, as well as software engineers who have worked on critical solutions at SpaceX, RTX and the US military, was recently selected to move on to the second phase.

“The general idea is that right now almost every ground vehicle used by the military has a person inside,” Boots told TechCrunch in a video interview. “And you can imagine that if you can just get the person out of the vehicle, that confers safety and tactical advantages.”

To remove the person, this means vehicles must autonomously navigate complex off-road terrain using only onboard sensors (primarily cameras, according to Boots) and calculations, without relying on maps, GPS or operators humans at a distance. This means that Overland’s software must understand the geometry of the ground, including things like vegetation and mud, at every step of the process, and how that affects vehicle dynamics.

“The terrain determines how the vehicle moves,” Boots said.

Overland’s technology “basically takes sensor data and builds a representation of the terrain as it goes,” Boots explained. Then the vehicle uses that digital representation “along with the goal it’s trying to reach, which may be several miles away, to try to find a route through the terrain toward that goal.”

“Part of the advantage of having an autonomous system is that when the system is charged, if you lose a communications link with that ground vehicle, it will continue to move towards its objective and attempt to complete the task until until the communication link is reestablished,” said Botte.

Today, most road travel relies on this telecommunications link to remote assistance, in part because the risk to other road users is higher. That’s why you’ll see Waymo and Cruise robotaxis set up on the streets of San Francisco, waiting for a remote operator to give them a nudge after you stop driving to meet a minimum safety requirement.

“Land-based military systems must often operate in dynamic, unstructured terrain. We believe that self-driving technology designed for well-defined streets and enclosed terrain will struggle there, and that it takes a very strong team to provide operationally relevant ground autonomy in these environments,” Chris Morales , associate in the defense technical team at Point72 Ventures. , told TechCrunch.

Potential of potential with off-road ADAS

“How can you actually help someone who may not be a 100% expert driver, but someone who wants to get off the beaten path and experience these more challenging conditions? » Sam Poirier, CEO of Potential, asked in a recent interview.

Potential’s core platform, called Terrain Intelligence, uses computer vision to help vehicles see, interpret and prepare for complex terrain and changing surface conditions ahead. Terrain Intelligence can read data from a single camera, rather than relying on additional sensors such as additional cameras, lidar and radar.

At the most basic level, Potential’s Off-Road ADAS alerts the driver to the presence of an impassable object ahead or the need to change to a better driving setting based on the new terrain.

“The second level is: can we really help automate parameter changes that are typically driver-assisted? » said Poirier. “Most vehicles have two-wheel drive, four-wheel drive, sand mode, mud mode, things like that. Ultimately, at this point, it’s up to the driver to switch between these modes…and the driver must understand when to use these different modes.

The final level of potential would involve using data from existing sensors, refining these parameters and pushing the limits of performance.

“There are things that assistive tools can do that an individual driver — no matter how much expertise you have — can’t do alone,” said former Jeep chief engineer and industry veteran Scott Kunselman. automotive and advisor to Potential. “Stability controls are a good example because to activate stability control you need independent control of the brakes. The driver has only one brake pedal and operates the entire braking system simultaneously. Whereas stability control can operate each wheel individually and that’s how you can produce, for example, the ability to compensate for a vehicle’s yaw.

By the way, yaw occurs when a vehicle’s weight shifts from its center of gravity to the right or left, which can cause it to derail or fishtail.

Potential said it is working with both Tier 1 suppliers and OEMs to license its software and integrate it directly into vehicles. Andreev suggests focusing on business relationships with Tier 1 suppliers rather than OEMs who are less likely to take a chance on a small startup.