Using AI to anticipate others’ behavior on the road

A motivating example of M2I. Top: Traditional marginal predictor often produces scene inconsistent trajectory predictions that collide with each other. Even for non-colliding predictions, it ignores the potential interaction between agent futures and may predict unrealistic behaviors. Bottom: Our proposed approach M2I predicts scene compliant trajectories by first identifying an influencer reactor pair in the scene. It then predicts marginal trajectories for the influencer and reactive trajectories for the reactor. Credit: Qiao Sun et al.

Humans may be one of the biggest roadblocks keeping fully autonomous vehicles off city streets.

If a robot is going to navigate a …

Be the first to comment

Leave a Reply

Your email address will not be published.