RetroNav improves robot navigation by learning from past mistakes
RetroNav: Scene-Grounded Retrospection and Retry for Self-Evolving Lifelong Navigation
Robotics
Summary
Robots often struggle to remember what they did wrong when moving through a space, which stops them from getting better over time. The authors created RetroNav, a system that helps robots keep track of both the layout of a place and their past navigation attempts. After each try, RetroNav decides if the robot should try again and guides it based on what went wrong before. This way, the robot learns from its mistakes and improves its navigation without extra training.
What this means in practice
- •For robotics engineers: Improve robot navigation by enabling robots to recall past navigation attempts and retry with evidence-based guidance.
- •For autonomous vehicle developers: Use scene-grounded retrospection to enhance route planning and error correction in repeated navigation tasks.
Authors
Diantao Tu, Yangdong Liu, Niannian Dai, Mengqi Rong, Shuhan Shen
Abstract
Existing robot navigation methods primarily focus on isolated tasks, overlooking the need to accumulate knowledge and experience across repeated navigation episodes. Although recent lifelong navigation methods have explored cross-task knowledge accumulation, they mainly focus on scene knowledge while neglecting historical navigation experience and its connection to scene evidence, limiting the ability to reflect on past mistakes and improve future decisions. To address this limitation, this paper presents RetroNav, a training-free self-evolving navigation framework that integrates navigation, memory accumulation, and failure reflection. Specifically, we develop a unified memory system that connects geometric and semantic scene knowledge with historical navigation experience, distinguishing objective observations from task-dependent interpretations. After each navigation attempt, an independent evaluator determines whether a retry is necessary. If so, a Retrospection Agent revisits historical decisions using accumulated scene knowledge and navigation records, generating an evidence-grounded retry brief to guide the next attempt. RetroNav achieves state-of-the-art performance across five navigation benchmarks. Our method will be open-sourced at https://github.com/3dv-casia/RetroNav .