Most self-driving cars use a combination of cameras, lidar, and radar to navigate crowded city streets, rural roads, and long stretches of highway. But WaveSense, a Somerville, Massachusetts-based startup that launched publicly earlier this year, contends that they’re not good enough.

The startup today announced a $3 million seed round backed by Rhapsody Venture Partners, Vas Ventures, NOMO Ventures, and “other leading VC firms.” CEO Tarik Bolat said the company will use this capital to commercialize its ground-penetrating radar (GPR) technology.

WaveSense, which came out of stealth in August, has its origins in the Massachusetts Institute of Technology’s Lincoln Laboratory. There, Bolat and others developed GPR systems designed to send very high frequency (VHF) electromagnetic pulses up to 10 feet below ground, where they reflect off of pipes, roots, rocks, dirt, and other underground features.

WaveSense’s tech builds a basemap from those objects and correlates it into a three-dimensional, GPS-tagged subterranean database. With an onboard computer running a complex algorithm, it’s able to iteratively narrow in on the car’s location as it moves.

The U.S. military used this technology to help military vehicles traverse regions with poor or nonexistant road markings. The first systems underwent testing in Afghanistan in 2013.

“The ground-penetrating radar technology … will accelerate the commercialization of self-driving vehicles and will significantly reduce civilian autonomous vehicle fatalities,” Byron Stanley, WaveSense cofounder and chief technology officer and a lead researcher on Lincoln Laboratory’s GPR program, said in a statement. “That mission has driven our work and our passion for a decade and is what propels us forward now.”

Unlike cameras and lidar, which are often stymied by inclement weather, WaveSense’s radars can penetrate rain, dust, fog, and snow. In 2016, Lincoln Labs researchers demonstrated that an SUV equipped with its system could stay within centimeters of its lane on a road freshly coated with snow.

“We’ve achieved 4cm lateral side-to-side accuracy at highway speeds and 6cm lateral accuracy in snowstorms in the middle of the night,” Bolat told VentureBeat last month. “I don’t believe any of the autonomous vehicle companies can lay claim to this.”

It’s impressive, but WaveSense has its work cut out for it — achieving the sort of localized tracking Bolat describes would require scanning each road individually. Still, the pitch was enough to convince several automotive partners to conduct pilots, and Bolat said more are on the roadmap.

“[Our] technology radically improves the safety of self-driving vehicles in all conditions and provides the confidence and reliability our sector must demonstrate in order to earn the public’s trust,” he said.

Sign up for Funding Daily: Get the latest news in your inbox every weekday.