Open-loop Planning on Bench2Drive (test)
0.73Avg L2 Error (m)UniAD-Base
Evaluation Results
| Method | Links | |
|---|---|---|
| UniAD-BaseInput=Ego State + 6 Cameras, Expert Feature Distillation=false2025.04 | 0.73 | |
| UniAD-TinyInput=Ego State + 6 Cameras, Expert Feature Distillation=false2025.04 | 0.8 | |
| FUMPInput=Ego State + 6 Cameras, Expert Feature Distillation=false2025.04 | 0.8 | |
| VADInput=Ego State + 6 Cameras, Expert Feature Distillation=false2025.04 | 0.91 | |
| Think2TwiceInput=Ego State + 6 Cameras, Expert Feature Distillation=true2025.04 | 0.95 | |
| DriveAdapterInput=Ego State + 6 Cameras, Expert Feature Distillation=true2025.04 | 1.01 | |
| DiffADInput=Ego State + 6 Cameras, Expert Feature Distillation=true2025.04 | 1.55 | |
| AD-MLPInput=Ego State, Expert Feature Distillation=false2025.04 | 3.64 |