Autonomous Driving on Bench2Drive 220 short segment scenarios
82.37Driving ScoreExpert
Evaluation results
| Method | Links | ||
|---|---|---|---|
| ExpertInput=Privileged2024.06 | 82.37 | 84.29 | |
| ExpertInput=Privileged2025.03 | 82.37 | 84.29 | |
| VL (on failure)Input=Privileged2024.06 | 77.3 | 72.25 | |
| ExpertInput=Camera2024.06 | 75.82 | 82.72 | |
| ExpertInput=Sensor2025.03 | 75.82 | 82.72 | |
| IL + RLInput=Privileged2024.06 | 74.57 | 60.41 | |
| VL (on failure)Input=Camera2024.06 | 73.29 | 65.44 | |
| CAPSInput=Privileged2025.03 | 68.91 | 56.97 | |
| CAPSInput=Sensor2025.03 | 66.76 | 52.87 | |
| DriveAdapterInput=Camera2024.06 | 64.22 | 33.08 | |
| DriveAdapterInput=Sensor2025.03 | 64.22 | 33.08 | |
| ThinkTwiceInput=Camera2024.06 | 62.44 | 31.23 | |
| ThinkTwiceInput=Sensor2025.03 | 62.44 | 31.23 | |
| IL with Uniform SamplingInput=Privileged2025.03 | 62.26 | 54.16 | |
| WoTETraj. Eval.=Model-based2025.04 | 61.71 | 31.36 | |
| VL (on expert)Input=Privileged2024.06 | 60.01 | 49.23 | |
| ILInput=Privileged2024.06 | 59.96 | 48.69 | |
| TCP-trajInput=Camera2024.06 | 59.9 | 30 | |
| TCP2025.04 | 59.9 | 30 | |
| TCP-trajInput=Sensor2025.03 | 59.9 | 30 | |
| IL with Uniform SamplingInput=Sensor2025.03 | 59.49 | 48.16 | |
| UniAD-BaseInput=Camera2024.06 | 45.81 | 16.36 | |
| UniADTraj. Eval.=Rule-based2025.04 | 45.81 | 16.36 | |
| UniAD-BaseInput=Sensor2025.03 | 45.81 | 16.36 | |
| VADInput=Camera2024.06 | 42.35 | 15 | |
| VADTraj. Eval.=Rule-based2025.04 | 42.35 | 15 | |
| VADInput=Sensor2025.03 | 42.35 | 15 | |
| AD-MLPInput=Camera2024.06 | 18.05 | 0 | |
| AD-MLP2025.04 | 18.05 | 0 | |
| AD-MLPInput=Sensor2025.03 | 18.05 | 0 |