Lift on Simulation
100Success RateObject-Centric Residual RL
Evaluation Results
| Method | Links | |
|---|---|---|
| Object-Centric Residual RLPolicy Configuration=+Res.2026.06 | 100 | |
| MVTTask type=Reinforcement Learning (PPO), Training timesteps=3 × 10^6, Evaluation=Average of 5 random seeds2026.02 | 97.9 | |
| ViTaSTask type=Reinforcement Learning (PPO), Training timesteps=3 × 10^6, Evaluation=Average of 5 random seeds2026.02 | 97.5 | |
| RLDC2026.06 | 96 | |
| RLDCSeparate Head=false2026.06 | 93 | |
| RLDCState Pre-training=false2026.06 | 85 | |
| ConcatTask type=Reinforcement Learning (PPO), Training timesteps=3 × 10^6, Evaluation=Average of 5 random seeds2026.02 | 76.7 | |
| PoETask type=Reinforcement Learning (PPO), Training timesteps=3 × 10^6, Evaluation=Average of 5 random seeds2026.02 | 71.9 | |
| VTTTask type=Reinforcement Learning (PPO), Training timesteps=3 × 10^6, Evaluation=Average of 5 random seeds2026.02 | 70.4 | |
| DP32026.06 | 70.3 | |
| CVTTask type=Reinforcement Learning (PPO), Training timesteps=3 × 10^6, Evaluation=Average of 5 random seeds2026.02 | 70.2 | |
| iDP32026.06 | 65 | |
| GR00T-N1.5Policy Configuration=Base2026.06 | 21.5 | |
| M3LTask type=Reinforcement Learning (PPO), Training timesteps=3 × 10^6, Evaluation=Average of 5 random seeds2026.02 | 20.6 |