Self-driving even when roads are covered with snow: MIT develops new navigation system

Typically, self-driving cars rely on a combination of cameras and other sensors to "see and confirm" their position on the road. But what should you do if the road markings are covered in snow? MIT has developed a system that allows cars to see snow-covered roads. The system uses experimental ground-penetrating radar positioning (LGPR) technology to locate TerraVision-equipped vehicles by sending radio waves to the surface, creating a digital fingerprint underground. Because radio waves can penetrate three meters underground and reflect off rocks, tree roots, soil and pipes. The reflected signals will be used to build a 3D map of underground conditions. Stitching such images together creates a complete 3D fingerprint that any vehicle equipped with LGPR technology can use to determine its location. Additionally, creating complete maps above and below ground increases the likelihood that autonomous vehicles can locate and navigate in any conditions. In closed-course testing, the MIT team has been able to accurately locate snow-covered road signs with an error margin of only about 1 inch (25 millimeters). Although the setup has so far only been tested at low speeds on country roads, the researchers believe it should also be usable at high speeds on motorways. However, it's not meant to replace the car's existing vision sensors, but to enhance them. A paper on the research, led by Professor Daniela Rus and doctoral student Teddy Ort, will be published later this month in IEEE Robotics and Automation Letters.