SOMERVILLE, Mass – While many companies are attacking the problem of keeping autonomous vehicles on the road through technologies such as advanced cameras, lidar, and GPS, a startup called WaveSense has taken the issue underground – literally.
The company is developing sensors with ground-penetrating radar (GPR) that uses ultra-wideband radar technology to create a map and track the subsurface beneath the road, providing reliable and safe navigation for autonomous vehicles and advanced driver assistance systems (ADAS). By using the GPR technology, vehicles can avoid issues related to road conditions and weather such as snow, heavy rain, fog, or poor lane markings.
The technology was originally developed for military applications through the MIT Lincoln Laboratory, and the team realized they could commercialize it for vehicles beyond military use cases, said Byron Stanley, co-founder and CTO at WaveSense. First deployed in Afghanistan in 2013, the technology allowed military vehicles to stay on previously mapped routes by matching radar measurements with maps of subterranean geology, the company said. This enabled precise navigation of 9-ton military vehicles despite unmarked lanes and poor visibility from sand and dust. With success there, the team started looking at commercialization opportunities.
“The team spent about five to six years in looking at the dual use of this technology, both in the military and commercial spaces,” said Stanley. “[We] determined there was a real possibility of this having a significant impact on the safety metrics of existing autonomous vehicle fleets.”
The company said every road in the world has a unique subsurface signature. The WaveSense technology can create a map of those subsurface signatures, which are then unaffected by changes happening above the road, such as snow, rain, fog, etc.
“It’s the most reliable way to do high-precision positioning that exists,” said Tarik Bolat, co-founder and CEO at WaveSense. “Some of the technologies out there like lidar or camera and GPS in some cases can be very, very precise in conditions that suit them. But obviously for an autonomous vehicle that’s ranging over all sorts of things in areas where conditions are changing, in both subtle and unsubtle ways, it needs to be very, very reliable.”
“A lot of the other sensors to date for the most part have tried to replicate how humans drive, using things like image classification or HD mapping,” added Stanley. “You’re sort of recognizing features that you know, which is the same thing you do when you drive home. You know where you’re going to make your left or your right, because you see the edge where the tree is – the tree is always there. But one of the shifts in perspective WaveSense is driving is that there’s actually more data to be leveraged, and you need to leverage that to make something that’s much safer and more reliable than a human driver. It’s just a different domain that’s outside of the typical kind of grooves of thinking for how to do autonomy.”
In the autonomous vehicle space, the company is looking to complement existing lidar and camera systems with mapping and localization features. “What people in Phoenix and San Francisco are trying to do, and in most people’s estimation, the cars will have lidar, camera and WaveSense on it, in order to get to the level of safety to drive adoption and have the regulators give the green light,” said Bolat.
Benefits beyond self-driving
But the company’s technology can also be used in non-autonomous passenger vehicles through their driver assistance systems, in cases where automakers do not want to install more expensive lidar for those applications. For example, a passenger vehicle with a camera, WaveSense and GPS can utilize applications such as driver lane alignment, and even self-driving valet parking in a parking garage, the company said.
In the case of a parking garage, the system would work in conjunction with cameras and GPS (if available) to have the user leave the car at the front of the building, and the car would self-park in the garage, with the user able to retrieve the car later at the push of a button through a mobile app. “That’s a feature that sort of surprised us when we first started out,” said Bolat, “but that’s sort of the broad feature being developed across OEMs.”
Beyond vehicles, additional markets for the technology includes warehousing, construction, utilities, mining, agriculture, sidewalk delivery, and even railways. For example, the system can provide accurate positioning for trains when in tunnels and valleys, or when differential corrections are not available. The system can also assist with automatic track inspection, checking for erosion and weaknesses in the underlying terrain, as well as rail tie health.
WaveSense raised $3 million in seed funding in September 2018, and currently has eight employees operating out of a co-working space in Somerville. While much of 2019 was spent developing pilot programs, Bolat said the company is looking to raise more funds for its next round of capital, along with making some partnership announcements as well as selling the systems to OEMs and their suppliers.