Constrained Reinforcement Learning on Safety Gymnasium CarCircle2 (eval)
16RewardFOCOPS
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| FOCOPSKnowledge Level=Oracle2025.04 | 16 | 29.2 | — | |
| CUPKnowledge Level=Oracle2025.04 | 15.4 | 28.3 | — | |
| TraCeSKnowledge Level=Zero Knowledge2025.04 | 15.2 | 11.9 | 4,341 | |
| TRPO-SauteKnowledge Level=Oracle2025.04 | 12.8 | 35.1 | — |