Researchers released PRM-as-a-Judge 1.5, an updated toolkit that evaluates robotic manipulation policies by converting video rollouts into fine-grained progress metrics rather than relying on binary success/failure labels. Version 1.5 adds three new metrics covering failure-side progress, post-drawdown recovery, and success-side execution quality, and introduces RoboPulse++ for testing the reliability of process reward models. The team released the assessment suite openly with benchmarks, metric implementations, and visualization tools.
