ResearchBenchmarksRobot Manipulation Success Analysis on PACK, CAP, USB, CASE 8 task-base strataFollow18Residual SumRPRO16.8224.78532.7540.715Jun 3, 2026Evaluation ResultsMethodMethodLinksResidual SumOR_MH Score95% Confidence Interval (Lower Bound)Chi-Squared Statistic (CMH)P-Value (1-sided)RPROComparison Baseline=PI...Comparison Baseline=PI0.6*2026.06182.241.4513.889.7RPROComparison Baseline=TPOComparison Baseline=TPO2026.06192.331.5115.234.8RPROComparison Baseline=DA...Comparison Baseline=DAgger-Buffered2026.06272.921.9226.81.1RPROComparison Baseline=DA...Comparison Baseline=DAgger2026.0647.54.583.0663.1310