Autonomous Vehicles Can See Through Rain and Snow Using Localized Ground-Penetrating Radar
MIT CSAIL's new system sends electromagnetic pulses underground that measure that area’s specific combination of soil, rocks, and roots.
Autonomous vehicles rely on LIDAR, cameras, and sensors to map their position while traveling on roads and highways. The technology works well in nominal conditions but can falter during bad weather conditions such as rain, fog, or snow. When it snows, for example, cameras can no longer see road markings and traffic signs, while LIDAR has trouble mapping obstacles when it rains or snows. To get around those issues, a team from MIT’s Computer Science and Artificial Intelligence Lab (CSAIL) has developed an unconventional solution using ground-penetrating radar (GPR).
The GPR system sends electromagnetic pulses underground to measure an area’s specific combination of soil, rock, and roots, which creates a unique fingerprint that the vehicle can use to localize its location while on the road. The process isn’t done in real-time, but rather mapped first, and then uploaded to the vehicle once mapping has been completed. According to team leader and Ph.D. student Teddy Ort, “If you or I grabbed a shovel and dug it into the ground, all we’re going to see is a bunch of dirt. But LGPR can quantify the specific elements there and compare that to the map it’s already created, so that it knows exactly where it is, without needing cameras or lasers.”
As it stands, the CSAIL team has only tested the GPR system at low speeds on a track of country road but believes with more development; it could be utilized for highways and other high-speed roads. They also acknowledge that their system doesn’t work very well in rainy conditions, as water seeps underground, affecting the measurements, and will have to be paired with other technology (sensors, lasers, etc.) because it can’t detect road hazards.