Reproduce “Crowd-Aware Robot Navigation for Pedestrians with Multiple Collision Avoidance Strategies via Map-based Deep Reinforcement Learning” 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
Shunyi Yao1, Guangda Chen, Quecheng Qiu, Jun Ma, Xiaoping Chen, Jianmin Ji
The authors propose a reinforcement-learning navigation policy that combines an egocentric sensor map with a three-channel pedestrian map encoding locations and velocities, plus the robot’s target pose. It is trained with pedestrians following SFM and ORCA. In random SFM simulation, PPO-PSD achieved a 0.968 success rate versus 0.840 for sensor-map-only PPO-SD, without establishing safety or generalization to broader human behavior.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.