Reproduce “Learning to swim in potential flow” 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
Yusheng Jiao, Feng Ling, Sina Heydari, Nicolas Heess, Josh Merel, Eva Kanso
The authors train a feedback policy for a simulated three-link fish to reach targets by changing body shape, then use shape-space geometry to interpret its movements. They compare policies with and without modeled-drift observations. At drift magnitude 0.15, the drift-aware policy reached both tested bearings across tested drift directions, while the naive policy failed when drift opposed the target; the result is limited to simulation conditions.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.