Reproduce “Positive Safety Margins from Artificial Potential Field Reward Augmentation for Size-Robust Obstacle Avoidance under Partial Observability” 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
Ziyi Hao
The authors train a simulated Panda reaching policy with an artificial potential field (APF) reward that penalizes proximity to pillars, although the policy observes neither pillar positions nor radius. This teaches a larger clearance buffer rather than size adaptation. At a test radius of 0.055 m, safe success is 89.4% versus 9.5% without APF, but collisions remain.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.