Self-driving cars learn to cope with bad weather

0

A team at the MIT CSAIL have developed a self-driving car that can cope with adverse weather conditions.

Most autonomous vehicles require localization, which is typically provided by a previously recorded map. Systems including lidar are sensitive to weather including fog and snow so the team at the Massachusetts Institute of Technology Computer Science and Artificial Intelligence Lab (MIT CSAIL) used Ground Penetrating Radar for localization.

The system was tested over 17km both in manual and autonomous driving modes in clear weather, rain and snow. The team found that the system permits safe driving in inclement weather including rain, snow and complete darkness.

Further research will look at modeling temperature and humidity induced reflection changes, and global localization techniques for initialization in the absence of prior GPS.

Share.

About Author

Comments are closed.