ResearchBenchmarksRobotic Task Execution on LIBERO 5-suite episode-weighted aggregateFollow52Success RatePOTR49.450.07550.7551.425May 23, 2026Evaluation ResultsMethodMethodLinksSuccess RateStepsL2 Error (Mean)L2 Error (Max)Accuracy ScoreJerkPOTR2026.0552289.30.0641.122.2264.22Naive2026.0549.7294.60.2241.9212.9895.483RTC2026.0549.5293.50.0831.4462.5184.623