VTM-Nav: Hierarchical Visual-Topological Memory for Cross-Episode Object-Goal Navigation

2026-07-15

AI Summary

This work addresses the challenge of effectively reusing self-acquired scene experiences across episodes in object-goal navigation without updating model parameters. The authors propose a training-free vision-and-language navigation framework that introduces, for the first time, a persistent hierarchical Visual-Topological Memory (VTM). This memory organizes and retrieves past experiences at both room and object levels, providing soft guidance only when the stored memory aligns with current observations. A conservative action execution mechanism is further integrated to mitigate oscillation and premature stopping. Experimental results demonstrate that the proposed method significantly outperforms an enhanced WMNav baseline across three standard benchmarks—HM3D v0.1, HM3D v0.2, and MP3D—achieving efficient and robust cross-episode experience reuse under fixed model parameters.

Abstract

Object-goal navigation requires an embodied agent to locate and reach an instance of a specified object category in an indoor environment. Recent training-free approaches leverage vision-language models (VLMs) for open-vocabulary semantic reasoning, but are typically evaluated under an episodic protocol that resets all scene-specific state after each episode. We introduce Cross-Episode Object-Goal Navigation, in which an agent repeatedly operates in the same scene, retains only self-acquired experience, and keeps its model parameters fixed. To support experience reuse, we present \method, a training-free VLM navigation framework with a persistent hierarchical Visual-Topological Memory (VTM). The VTM organizes scene knowledge at room and object levels and retrieves relevant experience through coarse-to-fine matching, providing memory as soft guidance only when it agrees with current observations. A conservative execution guard further mitigates oscillations, blocked motions, and premature stopping. Under a controlled same-scene protocol, we evaluate \method{} on three benchmarks, HM3D v0.1, HM3D v0.2, and MP3D, and compare it with a strengthened WMNav baseline augmented with cross-episode textual memory, while keeping the VLM backbone and action pipeline identical. \method{} achieves the best performance across all three benchmarks, demonstrating the effectiveness and robustness of structured visual-topological experience reuse across datasets.

Problem

• object-goal navigation
• cross-episode
• embodied AI
• experience reuse
• indoor environment

Innovation

• Visual-Topological Memory
• Cross-Episode Navigation
• Training-Free Navigation
• Experience Reuse
• Hierarchical Memory