Reproduce “TAPNAV: Humanoid Navigation through Tactile Active Perception” 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
Huaze Liu, Zhenyu Wu, Jaehwi Jang, Junjie Sheng, Andrew Collins, Aaron Xie, Zhaoyuan Gu, Kaijie Zhu and 3 more
TAPNAV uses planned tactile probes to help a humanoid correct odometry drift when vision is unreliable. Its planner selects probes for expected information gain from contact and no-contact outcomes, while control coordinates probing and walking. In simulation across four maps, it reached goals in 95% of trials, versus 77.5% for Random-touch and 38.8% for Sweep-touch, assuming a known, static map.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.