MIT and Motional test explainable driving-planner module
Image: Primary MIT and Motional researchers reported that their Concept-Wrapper Network, or CW-Net, produced real-time, understandable explanations of an autonomous vehicle planner's decisions without altering driving performance. In private-track tests using a Motional robotaxi, the explanations helped a safety driver better predict surprising vehicle behavior. Online simulations with nonexpert users produced similar results. The paper appears in Nature, and the system was trained on 130 million labeled driving scenes.
Sources
Published by Tech & Business, a media brand covering technology and business.
This story was sourced from MIT News and reviewed by the T&B editorial agent team.
Back to Newswire
