Autonomous Driving Planning on NAVSIM v2 (test)
100NCHuman
Evaluation Results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| HumanInput=—2026.04 | 100 | 100 | 87.4 | 100 | 98.1 | 100 | 99.8 | 100 | 90.1 | 90.3 | |
| Human Agent2026.05 | 100 | 100 | 87.4 | 100 | 98.1 | 100 | 99.8 | 100 | 90.1 | 90.3 | |
| SafeDrive2026.05 | 99.5 | 99 | 88.6 | 98.9 | 98.2 | 99.9 | 99.7 | 97.5 | 81.9 | 87.5 | |
| Vegabest-of-N (N=6) strategy=true2026.03 | 99.2 | 96.6 | 87.5 | 98.7 | 98.4 | 99.9 | 99.5 | 97.4 | 84.5 | 89.4 | |
| Vega2026.03 | 98.9 | 95.3 | 87 | 98.4 | 98.3 | 99.9 | 99.4 | 96.5 | 76.3 | 86.9 | |
| EvaDriveInput=C2026.04 | 98.8 | 98.5 | 96.6 | 98.4 | 97.8 | 99.8 | 98.9 | 94.3 | 55.9 | 86.3 | |
| IDOL2026.05 | 98.8 | 97.6 | 87.1 | 98.3 | 98.3 | 99.8 | 99.5 | 96.3 | 85.5 | 89.6 | |
| iPad2026.05 | 98.7 | 98 | 86.6 | 98.3 | 98.3 | 99.8 | 98.9 | 97.2 | 74.6 | 85.8 | |
| SGDrive-2Bnumber of parameters=2B, strategy=two-stage SFT2026.01 | 98.6 | 94.3 | 86 | 97.9 | 98.3 | 99.9 | 99.5 | 96.1 | 85.9 | 86.2 | |
| SGDriveVLA/VLM-based=true, extra synthetic/pseudo-expert/QA-style data=true2026.05 | 98.6 | 94.3 | 86 | 97.9 | 98.3 | 99.9 | 99.5 | 96.1 | 85.9 | 86.2 | |
| DriveVLA-W02026.03 | 98.5 | 99.1 | 86.4 | 98.1 | 97.9 | 99.7 | 98 | 93.2 | 58.9 | 86.1 | |
| DiffRefinerInput=C2026.04 | 98.5 | 97.4 | 87.6 | 97.7 | 98.3 | 99.8 | 99.6 | 97.7 | 86.2 | 86.2 | |
| Hydra-MDP++2026.05 | 98.5 | 98.5 | 87.4 | 97.9 | 98.2 | 99.7 | 99.5 | 95.8 | 75.7 | 85.6 | |
| DriveVLA-W0VLA/VLM-based=true2026.05 | 98.5 | 99.1 | 86.4 | 98.1 | 97.9 | 99.7 | 98 | 93.2 | 58.9 | 86.1 | |
| DiffRefiner2026.05 | 98.5 | 97.4 | 87.6 | 97.7 | 98.3 | 99.8 | 99.6 | 97.7 | 86.2 | 86.2 | |
| DriveSuprim2026.05 | 98.4 | 98.6 | 90.5 | 97.8 | 98.3 | 99.8 | 99.6 | 97 | 78.6 | 87.1 | |
| Drive-JEPA2026.05 | 98.4 | 98.6 | 88.4 | 97.8 | 97.9 | 99.8 | 99.1 | 97.6 | 84.8 | 87.8 | |
| GTRS-Dense + SimScaleextra synthetic/pseudo-expert/QA-style data=true, RL/Reward-guided/Reinforcement fine-tuning=true2026.05 | 98.4 | 98.8 | 87.9 | 98.1 | 97.6 | 99.9 | 99.4 | 96.4 | 58.8 | 84.6 | |
| ARTEMIS2026.01 | 98.3 | 95.1 | 81.5 | 97.4 | 100 | 99.8 | 98.6 | 96.5 | 98.3 | 83.1 | |
| ReCogDrive-8Bnumber of parameters=8B2026.01 | 98.3 | 95.2 | 87.1 | 97.5 | 98.3 | 99.8 | 99.5 | 96.6 | 86.5 | 83.6 | |
| ARTEMIS2026.03 | 98.3 | 95.1 | 81.5 | 97.4 | 98.3 | 99.8 | 98.6 | 96.5 | — | 83.1 | |
| ReCogDriveVLA/VLM-based=true, RL/Reward-guided/Reinforcement fine-tuning=true2026.05 | 98.3 | 95.2 | 87.1 | 97.5 | 98.3 | 99.8 | 99.5 | 96.6 | 86.5 | 83.6 | |
| MeanFuser2026.05 | 98.3 | 97.2 | 87.6 | 97.4 | 98.3 | 99.8 | 99.6 | 97.3 | 88.2 | 89.5 | |
| DiffusionDrive2026.03 | 98.2 | 95.9 | 87.5 | 97.3 | 98.3 | 99.8 | 99.4 | 96.8 | 87.7 | 84.5 | |
| HADInput=C+L2026.04 | 98.2 | 97.3 | 87.4 | 97.5 | 98.3 | 99.8 | 99.2 | 95.2 | 86.2 | 88.6 | |
| DiffusionDrive2026.05 | 98.2 | 95.9 | 87.5 | 97.3 | 98.3 | 99.8 | 99.4 | 96.8 | 87.7 | 84.5 | |
| HAD-LInput=C2026.04 | 98.1 | 97.2 | 87.2 | 97.3 | 98.3 | 99.8 | 99.3 | 95.3 | 86.4 | 88.5 | |
| DiffusionDrive2026.01 | 98 | 96 | 87.7 | 97.1 | 98.3 | 99.8 | 99.5 | 97.2 | 87.6 | 84.3 | |
| VADv22026.05 | 98 | 98.3 | 87.1 | 96.8 | 98.3 | 99.8 | 99.4 | 95.2 | 88.1 | 85.8 | |
| Hydra-MDP++2026.01 | 97.9 | 96.5 | 79.2 | 93.4 | 100 | 100 | 98.9 | 67.2 | 97.7 | 80.6 | |
| World4Drive2026.05 | 97.8 | 96.3 | 88.3 | 97.1 | 98 | 99.8 | 99.4 | 97.7 | 53.9 | 84.8 | |
| WorldRFTRL/Reward-guided/Reinforcement fine-tuning=true2026.05 | 97.8 | 96.5 | 88.5 | 97 | 98.1 | 99.8 | 99.5 | 97.4 | 69.1 | 86.7 | |
| Transfuser2026.01 | 97.7 | 92.8 | 79.2 | 92.8 | 100 | 99.9 | 98.3 | 67.6 | 95.3 | 77.8 | |
| DiffusionDriveV2Input=C+L2026.04 | 97.7 | 96.6 | 88.9 | 97.2 | 97.8 | 99.8 | 99.2 | 96 | 91 | 85.5 | |
| DiffusionDriveV2RL/Reward-guided/Reinforcement fine-tuning=true2026.05 | 97.7 | 96.6 | 88.9 | 97.2 | 97.8 | 99.8 | 99.2 | 96 | 91 | 85.5 | |
| Hydra-MDP2026.01 | 97.5 | 96.3 | 80.1 | 93 | 100 | 99.9 | 98.3 | 65.5 | 97.4 | 79.8 | |
| DriveSuprim2026.03 | 97.5 | 96.5 | 88.4 | 96.6 | 98.3 | 99.6 | 99.4 | 95.5 | 77 | 83.1 | |
| DriveSuprimInput=C2026.04 | 97.5 | 96.5 | 88.4 | 96.6 | 98.3 | 99.6 | 99.4 | 95.5 | 77 | 83.1 | |
| DriveSuprim2026.05 | 97.5 | 96.5 | 88.4 | 96.6 | 98.3 | 99.6 | 99.4 | 95.5 | 77 | 83.1 | |
| VADv22026.01 | 97.3 | 91.7 | 77.6 | 92.7 | 100 | 99.9 | 98.2 | 66 | 97.4 | 76.6 | |
| DAWN2026.05 | 97.3 | 92 | 87.4 | 96.6 | 98.3 | 99.7 | 99.1 | 96 | 85.5 | 83.2 | |
| HydraMDP++2026.03 | 97.2 | 97.5 | 83.1 | 96.5 | 98.2 | 99.6 | 99.4 | 94.4 | 70.9 | 81.4 | |
| HydraMDP++Input=C2026.04 | 97.2 | 97.5 | 83.1 | 96.5 | 98.2 | 99.6 | 99.4 | 94.4 | 70.9 | 81.4 | |
| Hydra-MDP++2026.05 | 97.2 | 97.5 | 83.1 | 96.5 | 98.2 | 99.6 | 99.4 | 94.4 | 70.9 | 81.4 | |
| Epona2026.05 | 97.1 | 95.7 | 88.6 | 96.3 | 98 | 99.7 | 99.3 | 97 | 67.8 | 85.1 | |
| TransFuser2026.03 | 96.9 | 89.9 | 87.1 | 95.4 | 98.3 | 99.7 | 97.8 | 92.7 | 87.2 | 76.7 | |
| TransfuserInput=C+L2026.04 | 96.9 | 89.9 | 87.1 | 95.4 | 98.3 | 99.7 | 97.8 | 92.7 | 87.2 | 76.7 | |
| Transfuser2026.05 | 96.9 | 89.9 | 87.1 | 95.4 | 98.3 | 99.7 | 97.8 | 92.7 | 87.2 | 76.7 | |
| TransFuser2026.05 | 96.9 | 89.9 | 87.1 | 95.4 | 98.3 | 99.7 | 97.8 | 92.7 | 87.2 | 76.7 | |
| Ego Status2026.03 | 93.1 | 77.9 | 86 | 91.5 | 98.3 | 99.6 | 92.7 | 89.4 | 85.4 | 64 | |
| Ego Status MLPInput=—2026.04 | 93.1 | 77.9 | 86 | 91.5 | 98.3 | 99.6 | 92.7 | 89.4 | 85.4 | 64 | |
| Ego Status MLP2026.05 | 93.1 | 77.9 | 86 | 91.5 | 98.3 | 99.6 | 92.7 | 89.4 | 85.4 | 64 |