Trajectory Prediction on nuScenes Targeted standardized evaluation (val)
0.65Average L2 Error (m)SparseDrive-S w/ RoCA trajectory prediction*
Evaluation results
| Method | Links | ||
|---|---|---|---|
| SparseDrive-S w/ RoCA trajectory prediction*Using Ego Status=true2025.06 | 0.65 | 0.25 | |
| PARA-Drive+Using Ego Status=true2025.06 | 0.7 | 0.24 | |
| SparseDrive-S w/ RoCA training regularizationUsing Ego Status=true2025.06 | 0.77 | 0.28 | |
| SparseDrive-SUsing Ego Status=true2025.06 | 0.85 | 0.31 | |
| VAD-Tiny w/ RoCA trajectory prediction*Using Ego Status=false2025.06 | 0.91 | 0.25 | |
| VAD-Tiny w/ RoCA training regularizationUsing Ego Status=false2025.06 | 0.99 | 0.28 | |
| PARA-DriveUsing Ego Status=false2025.06 | 1.08 | 0.24 | |
| AD-MLPUsing Ego Status=true2025.06 | 1.13 | 1.4 | |
| VAD-TinyUsing Ego Status=false2025.06 | 1.27 | 0.39 | |
| UniADUsing Ego Status=false2025.06 | 1.59 | 0.47 |