End-to-end planning on NavSim
98.2NCDiffusionDrive
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| DiffusionDriveTraining Approach=P-IL, Input=C & L2025.12 | 98.2 | 96.2 | 94.7 | 100 | 82.2 | 88.1 | |
| PARA-DriveTraining Approach=P-IL, Input=C2025.12 | 97.9 | 92.4 | 93 | 99.8 | 79.3 | 84 | |
| Hydra-MDPTraining Approach=P-IL, Input=C & L2025.12 | 97.9 | 91.7 | 92.9 | 100 | 77.6 | 83 | |
| EponaTraining Approach=SS-L, Input=C2025.12 | 97.9 | 95.1 | 93.8 | 99.9 | 80.4 | 86.2 | |
| UniADTraining Approach=P-IL, Input=C2025.12 | 97.8 | 91.9 | 92.9 | 100 | 78.8 | 83.4 | |
| WorldRFT (with RFT)Training Approach=SS-L & RL, Input=C2025.12 | 97.8 | 96.8 | 94 | 100 | 81.7 | 87.8 | |
| TransfuserTraining Approach=P-IL, Input=C & L2025.12 | 97.7 | 92.8 | 92.8 | 100 | 79.2 | 84 | |
| DriveXTraining Approach=SS-L, Input=C2025.12 | 97.5 | 94 | 93 | 100 | 79.7 | 84.5 | |
| WorldRFT (without RFT)Training Approach=SS-L, Input=C2025.12 | 97.5 | 96 | 94 | 100 | 80.9 | 87 | |
| LTFTraining Approach=P-IL, Input=C2025.12 | 97.4 | 92.8 | 92.4 | 100 | 79 | 83.8 | |
| World4DriveTraining Approach=SS-L, Input=C2025.12 | 97.4 | 94.3 | 92.8 | 100 | 79.9 | 85.1 | |
| VADv2Training Approach=P-IL, Input=C & L2025.12 | 97.2 | 89.1 | 91.6 | 100 | 76 | 80.9 | |
| LAW (Perception-free)Training Approach=SS-L, Input=C2025.12 | 96.4 | 95.4 | 88.7 | 99.9 | 81.7 | 84.6 |