Planning on nuPlan (test14)
90.69Overall ScoreDFP (ours)
Evaluation Results
| Method | Links | |
|---|---|---|
| DFP (ours)Type=Learning-based, Evaluation Protocol=Non-reactive (NR)2026.06 | 90.69 | |
| DFP (ours) †Type=Learning-based, Evaluation Protocol=Non-reactive (NR), components=Flow Matching + Mixture-of-Transformer2026.06 | 90.62 | |
| CoPlannerType=Learning-based, Evaluation Protocol=Non-reactive (NR)2026.06 | 90.31 | |
| Diffusion Planner *Type=Learning-based, Evaluation Protocol=Non-reactive (NR), reproduced=true2026.06 | 90.01 | |
| PLUTOType=Learning-based, Evaluation Protocol=Non-reactive (NR)2026.06 | 89.9 | |
| Flow PlannerType=Learning-based, Evaluation Protocol=Non-reactive (NR)2026.06 | 89.88 | |
| Diffusion PlannerType=Learning-based, Evaluation Protocol=Non-reactive (NR)2026.06 | 89.19 | |
| Log-replayType=Expert, Evaluation Protocol=Non-reactive (NR)2026.06 | 85.96 | |
| PlanTFType=Learning-based, Evaluation Protocol=Non-reactive (NR)2026.06 | 85.62 | |
| DFP (ours) †Type=Learning-based, Evaluation Protocol=Reactive (R), components=Flow Matching + Mixture-of-Transformer2026.06 | 83.59 | |
| Diffusion PlannerType=Learning-based, Evaluation Protocol=Reactive (R)2026.06 | 82.93 | |
| Flow PlannerType=Learning-based, Evaluation Protocol=Reactive (R)2026.06 | 82.93 | |
| DFP (ours)Type=Learning-based, Evaluation Protocol=Reactive (R)2026.06 | 81.96 | |
| Diffusion Planner *Type=Learning-based, Evaluation Protocol=Reactive (R), reproduced=true2026.06 | 79.61 | |
| PlanTFType=Learning-based, Evaluation Protocol=Reactive (R)2026.06 | 79.58 | |
| CoPlannerType=Learning-based, Evaluation Protocol=Reactive (R)2026.06 | 78.81 | |
| PLUTOType=Learning-based, Evaluation Protocol=Reactive (R)2026.06 | 78.62 | |
| Log-replayType=Expert, Evaluation Protocol=Reactive (R)2026.06 | 68.8 | |
| UrbanDriverType=Learning-based, Evaluation Protocol=Reactive (R)2026.06 | 67.15 | |
| PDM-Open*Type=Learning-based, Evaluation Protocol=Reactive (R), reproduced=true2026.06 | 57.23 | |
| PDM-Open*Type=Learning-based, Evaluation Protocol=Non-reactive (NR), reproduced=true2026.06 | 52.81 | |
| UrbanDriverType=Learning-based, Evaluation Protocol=Non-reactive (NR)2026.06 | 51.83 | |
| GameFormerType=Learning-based, Evaluation Protocol=Non-reactive (NR)2026.06 | 11.36 | |
| GameFormerType=Learning-based, Evaluation Protocol=Reactive (R)2026.06 | 9.31 |