CereVLA improves robot action accuracy with consequence-aware corrections
CereVLA: Cerebellum-Inspired Consequence-Aware Residual Governance for Efficient Vision-Language-Action Execution
Computer Vision and Pattern Recognition
Summary
Robots often make mistakes when carrying out complex tasks because they commit to action steps without checking their outcomes. The authors present CereVLA, a method that lets a robot make small corrections during tasks by predicting the future effects of those fixes. This method helps avoid bad outcomes by suppressing corrections that would cause problems. Tested on several benchmarks, CereVLA significantly boosts success rates and efficiency without needing to retrain the robot’s original policy.
What this means in practice
- •For robotics engineers: Improve robot task success and efficiency by integrating consequence-aware correction without retraining the base control policy.
- •For automation system developers: Enhance automated vision-language-action execution by selectively suppressing harmful corrective commands predicted through outcome evaluation.
Authors
Shuai Zeng, Yuxuan Liang, Hangmiao Hu, Fobao Zhou, Zixiang Wang, Wenxi Hong, Hang Zhao
Abstract
Action-chunked vision-language-action (VLA) policies improve inference efficiency, but limited feedback within committed action chunks can lead to accumulated execution errors. Residual adaptation can correct such deviations without retraining the VLA; however, existing corrections are typically optimized for reference-action consistency without explicitly considering their downstream consequences. To address this limitation, we present Cerebellum-Inspired Consequence-Aware Residual Governance (CereVLA), a unified framework that integrates lightweight residual refinement and predictive consequence evaluation into frozen VLA execution. Corrective actions are first generated by flow-based residual refinement, and their short- and interval-horizon consequences are then evaluated by a recurrent state-space model and a history-aware classifier. Residual corrections predicted to be unfavorable are selectively suppressed by a lightweight governor. Comparisons with state-of-the-art methods on LIBERO-10 and LIBERO-GOAL demonstrate the effectiveness of CereVLA. On SO-101, CereVLA increases task success from 57.5% to 90.0% and reduces mean control steps by 19.6% among successful trials, relative to the frozen SmolVLA baseline.