Robot Manipulation on LIBERO (Perturbation Robustness) benchmark leaderboard · SOTA2 Research

Robot Manipulation on LIBERO (Perturbation Robustness)

0.96Success Rate (Baseline)π0

Updated 5mo ago

Evaluation Results

Method
π02025.09	0.96	0.635	0.314	0.23	0.527	0.517	0.514	0.208	0.937	0.617	0.733	0.746	0.371	0.931	0.943	0.787	0.847	0.792	0.626
RobustVLA (w/o input regularization)2025.09	0.96	0.673	0.363	0.449	0.569	0.585	0.557	0.408	0.941	0.539	0.641	0.632	0.39	0.912	0.946	0.815	0.84	0.757	0.648
RobustVLA2025.09	0.955	0.698	0.36	0.423	0.587	0.597	0.938	0.938	0.955	0.695	0.944	0.927	0.408	0.942	0.956	0.913	0.939	0.802	0.766
BYOVLA2025.09	0.952	0.62	0.321	0.212	0.516	0.511	0.587	0.431	0.952	0.542	0.777	0.704	0.373	0.937	0.95	0.795	0.858	0.801	0.64
RobustVLA (w/o output regularization)2025.09	0.949	0.668	0.319	0.37	0.54	0.527	0.945	0.908	0.95	0.587	0.94	0.892	0.379	0.899	0.943	0.777	0.823	0.722	0.717
RobustVLA (w/o UCB)2025.09	0.949	0.695	0.359	0.413	0.576	0.623	0.843	0.785	0.952	0.623	0.707	0.623	0.39	0.912	0.924	0.771	0.836	0.746	0.693
DR2025.09	0.948	0.612	0.301	0.276	0.485	0.542	0.641	0.545	0.887	0.54	0.859	0.723	0.319	0.808	0.89	0.788	0.723	0.567	0.618