CalVLA: Shaping Native Confidence in Vision-Language-Action Policies with Outcome Supervision
cs.AI, cs.RO
Submitted: 2026-10-30
Updated: 2026-10-30
Comments: 8 pages, 2 figures. Accepted at the Science of Physical AI Safety (SPAIS) Workshop, CoRL 2026
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Sources
- Confidence Calibration in Vision-Language-Action Models
- Shifting Uncertainty to Critical Moments: Towards Reliable Uncertainty Quantification for VLA Model
- Temporal Difference Calibration in Sequential Tasks: Application to Vision-Language-Action Models
- SAFECAST: Robust Failure Detection for VLA Policies with Contrast-Set Training and Calibration
- Hide-and-Seek in Trajectories: Discovering Failure Signals for VLA Runtime Monitoring
- ActProbe: Action-Space Probe for Early Failure Detection of Generative Robot Policies
- INSIGHT: INference-time Sequence Introspection for Generating Help Triggers in Vision-Language-Action Models
- CycleVLA: Proactive Self-Correcting Vision-Language-Action Models via Subtask Backtracking and Minimum Bayes Risk Decoding
Related papers
- Multi-Objective Structured Pruning of LLMs for Latency and Model Size Optimization
- MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
- Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
- Embeddings for Preferences, Not Semantics
- Can We Trust Item Response Theory for AI Evaluation?
- Diagnosing Task Insensitivity in Language Agents