AI RESEARCH
TrajRAG: Retrieving Geometric-Semantic Experience for Zero-Shot Object Navigation
arXiv CS.CV
•
ArXi:2605.01700v1 Announce Type: new Existing zero-shot Object Goal Navigation (ObjectNa) methods often exploit commonsense knowledge from large language or vision-language models to guide navigation. However, such knowledge arises from internet-scale text rather than embodied 3D experience, and episodic observations collected during navigation are typically discarded, preventing the accumulation of lifelong experience.