DyGeoVLN: Infusing Dynamic Geometry Foundation Model into Vision-Language Navigation
This addresses the problem of VLN agents struggling with dynamic environments for applications like robotics and autonomous navigation, representing a significant advancement over static scene assumptions.
The paper tackles the challenge of Vision-Language Navigation (VLN) in dynamic, real-world scenarios by proposing DyGeoVLN, a framework that infuses a dynamic geometry foundation model and uses a token-pruning strategy, achieving state-of-the-art performance on multiple benchmarks.
Vision-language Navigation (VLN) requires an agent to understand visual observations and language instructions to navigate in unseen environments. Most existing approaches rely on static scene assumptions and struggle to generalize in dynamic, real-world scenarios. To address this challenge, we propose DyGeoVLN, a dynamic geometry-aware VLN framework. Our method infuses a dynamic geometry foundation model into the VLN framework through cross-branch feature fusion to enable explicit 3D spatial representation and visual-semantic reasoning. To efficiently compress historical token information in long-horizon, dynamic navigation, we further introduce a novel pose-free and adaptive-resolution token-pruning strategy. This strategy can remove spatio-temporal redundant tokens to reduce inference cost. Extensive experiments demonstrate that our approach achieves state-of-the-art performance on multiple benchmarks and exhibits strong robustness in real-world environments.