Test-Time Trajectory Optimization for Autonomous Driving
For autonomous driving researchers, TOAD offers a plug-and-play method to enhance existing planners without retraining, addressing the limitation of fixed candidate sets.
TOAD improves autonomous driving planning by using test-time trajectory optimization with the Cross-Entropy Method, achieving 94.7 PDMS on NAVSIM-v1 and 56.3 EPDMS on NAVSIM-v2 across six base planners.
End-to-end planners for autonomous driving typically generate a set of candidate trajectories, score each one, and return the highest-scoring candidate. However, the scorer is applied only after the proposals are generated and cannot influence the set of trajectories: a weak set of candidates limits planning performance regardless of the scorer's quality. We instead treat the scorer as a learned trajectory-level reward function and search for trajectories that maximize it. Our method, TOAD, runs the Cross-Entropy Method at test time, warm-started from the planner's proposals. It requires no retraining and is plug-and-play for existing planners. Across six base planners, TOAD improves results on NAVSIM-v1 (94.7 PDMS), NAVSIM-v2 (56.3 EPDMS), and the closed-loop HUGSIM benchmark. The code will be made publicly available via the project page: https://valeoai.github.io/TOAD/.