VISTA: Scale-Aware Visual Navigation via Action History Conditioning
For robotics researchers deploying vision-based navigation models, VISTA provides a robust solution to scaling-induced failures and visual ambiguity, enabling reliable zero-shot navigation across diverse environments.
VISTA addresses the vulnerability of vision navigation models to action scaling factors by conditioning on normalized action histories, and improves performance in visually repetitive environments using a DINOv3 encoder. It achieves 100% goal prediction accuracy and 95% average checkpoints crossed in zero-shot real-world deployment.
Vision Navigation Foundation Models (VNMs) promise end-to-end learned navigation policies capable of zero-shot deployment across diverse embodiments and environments. To maintain generality, many vision-based navigation models predict normalized actions. However, this normalization introduces a critical deployment vulnerability: applying different scaling factors to the same normalized trajectory alters its physical geometry, which degrades navigation performance and increases collision risks. We address this vulnerability by conditioning the model on normalized action histories alongside image observations, providing explicit context on the relationship between the model's predictions and the robot's actual physical displacement. Furthermore, current VNMs often struggle in visually repetitive environments that lack distinct features. To resolve this issue, we integrate a DINOv3 encoder, whose richer representations enable our model to capture both spatial and geometric dimensions between observations. VISTA generalizes robustly to out-of-distribution environments, achieving 100% goal prediction accuracy in zero-shot, real-world deployment in Outdoor, Forest and Office settings, and an average of 95% checkpoints crossed, demonstrating consistent path following in unseen environments.