Robot Policy Learning on RLBench
41.2Button Success RateToBo
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| ToBoBackbone=ViT-S/16, Pre-trained=Kinetics-400 [26]2025.07 | 41.2 | 52.3 | 42.2 | 35.4 | 37 | |
| RSPBackbone=ViT-S/16, Pre-trained=Kinetics-400 [26]2025.07 | 28.4 | 48 | 37.3 | 31.9 | 18.5 | |
| CropMAEBackbone=ViT-S/16, Pre-trained=Kinetics-400 [26]2025.07 | 26.9 | 16.6 | 37.5 | 33.2 | 20.6 | |
| DINOBackbone=ViT-S/16, Pre-trained=Kinetics-400 [26]2025.07 | 24.7 | 32 | 28.1 | 31.4 | 12.9 | |
| MoCo v3Backbone=ViT-S/16, Pre-trained=Kinetics-400 [26]2025.07 | 11.4 | 36.2 | 13.2 | 8.7 | 6.7 | |
| SimCLRBackbone=ViT-S/16, Pre-trained=Kinetics-400 [26]2025.07 | 7.4 | 34.6 | 5.8 | 11 | 5.2 | |
| MAEBackbone=ViT-S/16, Pre-trained=Kinetics-400 [26]2025.07 | 6.4 | 37.7 | 10 | 10 | 6.2 | |
| SiamMAEBackbone=ViT-S/16, Pre-trained=Kinetics-400 [26]2025.07 | 6.1 | 5.4 | 4 | 8.7 | 3.5 |