Search papers, labs, and topics across Lattice.
The paper introduces VLAConf, a one-class discriminative confidence framework for Vision-Language-Action (VLA) models that estimates task success likelihood. VLAConf leverages frozen pretrained VLA internal representations to directly estimate step-wise anomaly scores with a lightweight confidence head in a single forward pass, avoiding the resampling overhead of existing methods. Experiments on the LIBERO benchmark and real-robot experiments show VLAConf significantly improves confidence signal quality and inference efficiency compared to existing baselines.
VLAConf offers a computationally cheap, architecture-agnostic way to estimate task success confidence for robot manipulation by turning frozen VLA representations into anomaly scores.
Confidence estimation for Vision-Language-Action (VLA) models is essential for robots to perform manipulation tasks in the open world, providing crucial signals for risk-sensitive decision-making and failure anticipation. Existing confidence estimation methods typically rely on ensemble-based paradigms or action-token probabilities to predict the likelihood of task success. However, they still encounter challenges in computational efficiency and cross-architecture generalizability. These methods usually require repeated sampling, leading to inference inefficiency, and are restricted to VLA models with discrete action outputs, making them difficult to apply to continuous action spaces. To address this issue, we propose VLAConf, a one-class discriminative confidence framework. By leveraging frozen pretrained VLA internal representations, VLAConf directly estimates step-wise anomaly scores in a single forward pass using a lightweight confidence head, thereby eliminating the overhead of exhaustive resampling. We additionally use step-conditioned modeling to encode rollout-phase information along the manipulation trajectory. Experiments on the LIBERO benchmark demonstrate that VLAConf significantly improves the quality of the confidence signal constructed for post-hoc calibration, outperforming existing baselines by a large margin in inference efficiency. The effectiveness of VLAConf is further validated in real-robot experiments. To access the source code and supplementary videos, visit https://sites.google.com/view/vlaconf.
