Temporal Difference Calibration in Sequential Tasks: Application to Vision-Language-Action Models

arXiv:2604.20472v1 Announce Type: cross Abstract: Recent advances in vision-language-action (VLA) models for robotics have highlighted the importance of reliable uncertainty quantification in sequential tasks. However, assessing and improving calibration in such settings remains mostly unexplored, especially when only partial trajectories are observed. In this work, we formulate sequential calibration for episodic tasks, where task-success confidence is produced along an episode, while success is determined at the end of it. We introduce a sequential extension of the Brier score and show that, for binary outcomes, its risk minimizer coincides with the VLA policy's value function. This connection bridges uncertainty calibration and reinforcement learning, enabling the use of temporal-difference (TD) value estimation as a principled calibration mechanism over time. We empirically show that TD calibration improves performance relative to the state-of-the-art on simulated and real-robot data. Interestingly, we show that when calibrated using TD, the VLA's single-step action probabilities can yield competitive uncertainty estimates, in contrast to recent findings that employed different calibration techniques.

Related

• Jump-Start Reinforcement Learning with Vision-Language-Action Regularization
• Can Explicit Physical Feasibility Benefit VLA Learning? An Empirical Study
• AVA-VLA: Improving Vision-Language-Action models with Active Visual Attention
• Improving Semantic Uncertainty Quantification in Language Model Question-Answering via Token-Level Temperature Scaling
• The Illusion of Certainty: Decoupling Capability and Calibration in On-Policy Distillation

Source: arXiv cs.LG | 2026-04-23