Reproduce “Dynamics-Decoupled Trajectory Alignment for Sim-to-Real Transfer in Reinforcement Learning for Autonomous Driving” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Thomas Steinecker, Alexander Bienemann, Denis Trescher, Thorsten Luettel, Mirko Maehlisch

The method trains an RL driver in simulation, distills it into a trajectory predictor, and synchronizes a virtual car with a real vehicle following its plan. A Stanley controller steers laterally; feedback and virtual-car timing address longitudinal mismatch. On one paved, closed 1.9 km course, mean lateral and longitudinal errors were 2.9 cm and 6.8 cm; other conditions were not tested.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.