Do VLAs Know They Will Fail? Interpretable Success and Failure Patterns in Physical AI Systemswith Christian A. Rogers, Daniel S. Brown1 versionW_u3h3bj3hv1 · current