VLA-Corrector: Lightweight Detect-and-Correct Inference for Adaptive Action Horizon
VLA-Corrector is a new framework that improves the robustness of Vision-Language-Action models in contact-rich manipulation tasks. It achieves this by introducing a lightweight latent-space vision monitor that enables adaptive corrective replanning. The framework can be integrated into different VLA models without further retraining.
Read original ↗Sign in to upvoteSign in to save