WaveSense Uses Ground-Penetrating Radar to Help Driverless Cars See
Those of us keeping tabs on the mobility industry have heard a lot about the development of sensors and other computer vision technologies to help autonomous vehicles “see” and interpret the landscape around them. WaveSense, a startup spun out of MIT, is adding ground-penetrating radar to the mix, which the company says could increase navigation precision and safety.
Tarik Bolat, WaveSense’s CEO and co-founder, says that in order to travel safely, self-driving cars must be able to determine what’s around them and exactly where they are. “We answer that question comprehensively,” he says. (Although ground-penetrating radar has been used for decades to do things like measure glaciers’ depths, Bolat says the technology hasn’t been used to guide autonomous vehicles.)
WaveSense’s technology comes out of MIT’s Lincoln Lab—the same lab that produced Humatics, a startup commercializing millimeter-scale navigation tech for industrial and automotive applications—where it was originally developed for military use in Afghanistan.
Ground-penetrating radar complements other sensor systems already on the vehicle—GPS, lidar, radar, cameras—to help driverless vehicles stay precisely in their lanes even when lane markings aren’t visible due to snow, fog, rain, or wear and tear. (Above-ground sensors are often hampered by weather and other visibility issues, Bolat says.) By adding subsurface data to above-ground sensor inputs, autonomous vehicles get a clearer picture of where they are, he adds.
“We use ground-penetrating radar to create a map of the subsurface,” Bolat explains, describing it as a “subterranean fingerprint” that reaches up to 10 feet below the ground. Driverless cars can use the WaveSense system to orient themselves by getting a fix on stable underground features even if the above-ground sensors are obscured. Bolat says the system is accurate to within a few centimeters at standard highway speeds.
“We can also operate in areas with confusing or no lane markings, or in low-landmark areas with no trees or signs,” he says. “Right now [with above-ground sensors], those areas are rendered unnavigable. Our belief, through conversations we’ve had with people in the sector, is that surface-only maps and sensors are necessary but insufficient because of the dynamism of surface environments, creating inconsistencies.”
WaveSense’s technology sends an electromagnetic pulse into the ground and generates a basemap using reflections of underground features such as pipes, rocks, and even architectural remnants. While the vehicle is in motion, WaveSense continually scans the underground layers of soil and road bedding and compares those scans to the images in its onboard database to determine the exact vehicle position relative to the road. It’s an approach Bolat says is unique to the market, although there are groups, such as San Antonio, TX’s Southwest Research Institute, that are developing autonomous vehicle technology based on algorithms that process images of the road surface below the car.
Bolat says WaveSense, which launched in 2017, is currently working with industry partners to test and commercialize its technology. The company says it will provide underground maps as a service to its customers, or it will enable partners to make their own maps utilizing the WaveSense system. “I can’t disclose who they are, but we’re talking to large automakers, large tech companies, ride-hailing companies,” and more. Earlier this month, the three-person WaveSense team announced that it had raised $3 million in a venture round led by Cambridge, MA-based Rhapsody Venture Partners.
“The opportunity and challenges for autonomous vehicles are well documented,” said Carsten Boers, managing partner of Rhapsody Venture Partners, in a prepared statement. “WaveSense addresses an unsolved problem in a unique way: To place the vehicle securely in all conditions, it looks underneath the road. Doing so is necessary for safe autonomous driving and naturally is also an important safety feature for assisted driving. We’re very excited to help WaveSense bring this military-grade technology to the broader market.”
“We view WaveSense as an enabling technology to accelerate the deployment of autonomous vehicles into the larger market,” Bolat says. “In the near-term, it will be an after-market product sold to our [supplier] partners sometime in the next year. Over time, we’d like to transition away from after-market and sell WaveSense directly to automakers or fleet owners.”