RoboDojo introduces a unified benchmark for evaluating generalist robot manipulation policies in both simulation and the physical world. It contains 42 simulation tasks and 18 real-world tasks, covering generalization, memory, precision, long-horizon execution, and open-vocabulary instruction following. Its RoboDojo-RealEval system standardizes hardware, scene reset, protocols, deployment interfaces, and remote cloud access. Together with XPolicyLab, policies can be integrated once and evaluated across both settings. The authors report integrating 30 policies and establishing a public leaderboard, aiming to make manipulation evaluation broader, more reproducible, and more connected to deployment conditions.
No heat snapshots are available in the last 24 hours.