Reproduce “Ask to Evolve: A Unified Harness for Lifelong Embodied Navigation with Hybrid Action Space and Persistent Memory” 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

Kai Sheng, Liuyi Wang, Jinlong Li, Haojie Dai, Chengju Liu, Qijun Chen

EvoLife combines four movement modes, interaction with a guide or nearby person, and place-linked memory in a navigation loop. It stores descriptions, images, and dialogue with remembered locations, then retrieves them on revisits. In simulation, it reaches 100% success on four 100-episode benchmark sets; physical evidence is three successful tasks over two days, not broad reliability.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.