The authors extend a recurrent state-space world model (RSSM) with an intent inference module that predicts a human driver's discrete objectives from observed behavior.
Shared driving assistant infers human intent to align torque and avoid collisions
Toyota Research Institute and MIT researchers present Dream2Assist, an intent-conditioned world model for shared vehicle control, showing improved torque alignment and collision rates in human-subject studies.
Big Tech
Jonathan DeCastro · Thomas Balch · Emily Sumner · Deepak Gopinath · Laporsha Dees · Patricio Reyes Gomez · +6 more
Toyota Research Institute · Massachusetts Institute of Technology
Research Digest··2 min read
DeCastro et al. introduce Dream2Assist, a model-based reinforcement learning framework that infers a driver's discrete intentions and conditions assistive control on them.
The Daily Edition
The front page in your inbox each weekday morning: the lead, the picks, and why each one earned its place.