Abstract
Zero-Shot Object-Goal Navigation (ZS-OGN) requires embodied agents to explore and locate target objects without any prior training. To this end, recent methods leverage foundation models. However, they typically rely on static priors and lack adaptation, leading to repeated errors and costly trial and error. This paper proposes a self-evolving ZS-OGN framework that enables continuous test-time improvement.
Specifically, we build an agentic rule memory by extracting actionable knowledge from past trajectories. Then, we propose a retrieval strategy based on upper confidence bound, selecting effective rules by balancing semantic relevance and historical success. Additionally, we introduce a memory-guided preflection module that forecasts potential outcomes before action, reducing inefficient exploration. Extensive experiments show that our method outperforms existing zero-shot baselines, achieving a 10.1% improvement in success rate with fewer unnecessary steps.
Blogger's Review: EvolveNav's innovation lies in its self-evolving memory mechanism and proactive preflection module, significantly enhancing the efficiency of zero-shot navigation. This approach not only offers new insights for autonomous learning in agents but also lays a foundation for future research, making it a topic worthy of attention and deeper exploration.