Reproduce “Multi-task UNet architecture for end-to-end 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
Der-Hau Lee, Jinn-Liang Liu
Lee and Liu present an integrated lane-keeping pipeline combining camera-based predictions with path prediction and conventional steering and speed controllers. A shared image network predicts lane markings, heading, road-turn class, and lead-car distance class. In TORCS simulation on unseen Track 8, the residual multi-task model reports dynamic heading error of 0.0059 rad and lateral-offset error of 6.58 cm; this does not establish public-road safety.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.