Reproduce “NaviSTAR: Socially Aware Robot Navigation with Hybrid Spatio-Temporal Graph Transformer and Preference 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
Weizheng Wang, Ruiqi Wang, Le Mao, Byung-Cheol Min
NaviSTAR models crowd navigation by representing nearby interactions and agents’ movement histories with spatial and temporal transformer blocks, fusing them with a cross-modal transformer, and learning rewards from human comparisons of trajectory segments. In simulation, its social score is 98.9 versus 95.7 for NaviSRNN in the 10-human setting, but not highest at FOV 360° and people do not react.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.