Source: MIT CSAILResearchers at the Massachusetts Institute of Technology (MIT) Computer Science and Artificial Intelligence Lab (CSAIL) have developed a technique to enable autonomous vehicles to navigate through rain and snow.

The MIT CSAIL team developed a system called localizing ground penetrating radar (LGPR), which is a type of ground penetrating radar that emits electromagnetic pulses underground to gauge an area’s composition of soil, rocks and roots. The mapping technique assigns a unique fingerprint to the location based on the area’s composition of rocks, soil and roots that the autonomous vehicle will recognize, thereby enabling the self-driving vehicle to localize itself without having to exclusively rely on the cameras and sensors that enable a self-driving vehicle to see.

One of the hindrances to the safety of self-driving vehicles is their ability to navigate through snow and rain. Oftentimes, the autonomous vehicles’ vision is compromised by weather conditions that interfere with the automobile's lidar sensors and cameras, thereby obstructing the vehicle's view.

Although the LGPR is expected to help improve autonomous vehicle navigation through such conditions, the technology, according to the MIT CSAIL team, is far from ready. Likewise, the technology is only to be considered as a complement to lidar sensors and cameras because the ground penetrating technology does not detect anything in the self-driving vehicle’s path.

For now, the MIT CSAIL team is attempting to reduce the size of the LGPR system, which is currently a bulky 6 ft wide.

For more on the technology, watch the accompanying video that appears courtesy of MIT CSAIL or read the team's article detailing the technology, which appears in IEEE Robotics and Automation Letters.