Task Diversity and Scene Feasibility Evaluation on Robotics Task Generation (Evaluation Set)
106Number of TasksRoboGen
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| RoboGen2026.02 | 106 | 28.4 | 16.5 | 19.3 | |
| RLBench2026.02 | 106 | 31.7 | 20 | 37.5 | |
| AGT-World2026.02 | 102 | 80.6 | 37.6 | 44 | |
| Behavior-1002026.02 | 100 | 29.9 | 21 | 38.9 | |
| GenSim-V22026.02 | 70 | 37.8 | 28.8 | 71.7 | |
| Meta-World2026.02 | 50 | 32.2 | 26.3 | 51.7 | |
| ManiSkill22026.02 | 20 | 67.4 | 19.4 | 33.2 |