Reproduce “Place Cells as Multi-Scale Position Embeddings: Random Walk Transition Kernels for Path Planning” 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
Minglu Zhao, Dehong Xu, Deqian Kong, Wen-Hao Zhang, Ying Nian Wu
The authors present a multi-scale cognitive map in which nonnegative place-cell activity vectors represent how locations connect through movement over different time horizons. Vector overlap approximates random-walk transition probability, and a planner follows gradients in goal similarity to route around obstacles. In Four-Room simulations, it achieved 100% success and SPL 1.00, versus A*’s 1.08 ± 0.50; environments were simulated.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.