SOURCE-LINKED INTELLIGENCE

Breaking the Vision-Action Shortcut: Latent Interface Training for Generalizable Robotics Foundation Models

arXiv · AI, language, vision and robotics · article · Sep 11, 2026 · UTC

Robot foundation models achieve strong in-distribution performance but often degrade under visual distribution shifts. When learning to generate actions from pretrained visual representations, models may exploit task-irrelevant visual cues that correlate with demonstrated actions within the training distribution. Such vision-action shortcuts can undermine generalization when these correlations change under distribution shifts. Mitigating these shortcuts requires constraining how visual information is used for action generation while preserving task-relevant spatial information. We propose Late

Read original source ↗ Open in workspace

recordType

paper

region

Global