Closed-loop Planning on nuPlan 14 (val)
95NR ScoreCarPLAN
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| CarPLANType=Learning, Post-processing=True2026.03 | 95 | 93.8 | — | |
| FlowDrive*Type=Rule-based & Hybrid2025.09 | 94.81 | 92.96 | — | |
| Pluto-SoEType=Hybrid, Time (ms)=2592025.12 | 94.56 | 91.17 | — | |
| Diffusion-PlannerType=Learning, Post-processing=True2026.03 | 94.3 | 92.9 | — | |
| DiffusionType=Hybrid2025.12 | 94.26 | 92.9 | — | |
| Diffusion PlannerType=Rule-based & Hybrid, Refinement=Enabled2025.09 | 94.26 | 92.9 | — | |
| Diffusion Planner w/ refineType=Rule-based & Hybrid2026.02 | 94.26 | 92.9 | — | |
| Log-replayType=Expert2026.01 | 93.53 | 80.32 | — | |
| Log-replayType=Expert2026.01 | 93.53 | 80.32 | — | |
| Log-ReplayType=Rule2025.12 | 93.53 | 80.32 | — | |
| Log-replayType=Expert2025.09 | 93.53 | 80.32 | — | |
| Log-replayType=Expert2026.02 | 93.53 | 80.32 | — | |
| Log-replayType=Expert, Post-processing=True2026.03 | 93.5 | 75.9 | — | |
| PLUTOType=Learning, Post-processing=True2026.03 | 93.2 | 76.8 | — | |
| PLUTOType=Rule-based & Hybrid2025.09 | 92.88 | 89.88 | — | |
| PLUTOType=Rule-based & Hybrid2026.02 | 92.88 | 76.88 | — | |
| PDM-ClosedType=Rule2026.01 | 92.84 | 92.12 | — | |
| PDM-ClosedType=Rule2026.01 | 92.84 | 92.12 | — | |
| PDM-ClosedType=Rule, Time (ms)=1222025.12 | 92.84 | 92.13 | — | |
| PDM-ClosedType=Rule-based & Hybrid2025.09 | 92.84 | 92.12 | — | |
| PDM-ClosedType=Rule-based & Hybrid2026.02 | 92.84 | 92.12 | — | |
| PDM-ClosedType=Rule, Post-processing=True2026.03 | 92.8 | 92.1 | — | |
| PDM-HybridType=Learning, Post-processing=True2026.03 | 92.8 | 92.1 | — | |
| PlutoType=Hybrid, Time (ms)=2592025.12 | 92.79 | 89.77 | — | |
| PDM-HybridType=Rule-based & Hybrid2025.09 | 92.77 | 92.11 | — | |
| PDM-HybridType=Rule-based & Hybrid2026.02 | 92.77 | 92.11 | — | |
| FlowDriveType=Learning-based2025.09 | 91.21 | 85.37 | — | |
| RaScType=Learning, Post-processing=True2026.03 | 91 | 86 | — | |
| Pluto w/o-SoE1Type=Learning, Time (ms)=1072025.12 | 90.94 | 81.81 | — | |
| Flow PlannerType=Learning2026.01 | 90.43 | 83.31 | — | |
| Flow PlannerType=Learning2026.01 | 90.43 | 83.31 | — | |
| FlowDrive-Type=Learning-based2025.09 | 90.3 | 81.91 | — | |
| RAPiDType=Learning-based2026.02 | 90.19 | 81.84 | — | |
| Pluto w/o-SoE2Type=Learning, Time (ms)=1072025.12 | 90.04 | 82.36 | — | |
| PlannerRFTType=Learning2026.01 | 89.96 | 84.46 | — | |
| PlannerRFTType=Learning2026.01 | 89.96 | 84.46 | — | |
| Diffusion PlannerType=Learning2026.01 | 89.87 | 82.8 | — | |
| Diffusion PlannerDPMType=Learning, Sampler=10-step DPM-solver2026.01 | 89.87 | 82.8 | — | |
| Diffusion PlannerType=Learning-based2025.09 | 89.87 | 82.38 | — | |
| DiffusionPlannerType=Learning-based2026.02 | 89.87 | 82.8 | — | |
| Diffusion w/oType=Learning, Time (ms)=1592025.12 | 89.85 | 82.88 | — | |
| Diffusion PlannerDDIMType=Learning, Sampler=5-step DDIM2026.01 | 89.81 | 82.94 | — | |
| PLUTOType=Learning2026.01 | 88.89 | 78.11 | — | |
| PLUTOType=Learning2026.01 | 88.89 | 78.11 | — | |
| PLUTOType=Learning-based, Refinement=None2025.09 | 88.89 | 78.11 | — | |
| PLUTO w/o refineType=Learning-based, Pre-computed reference paths=true2026.02 | 88.89 | 78.11 | — | |
| Pluto w/oType=Learning, Time (ms)=1072025.12 | 88.78 | 78.01 | — | |
| planTF-SoEType=Learning, Time (ms)=782025.12 | 86.9 | 79.7 | — | |
| planTFType=Learning, Time (ms)=782025.12 | 84.69 | 76.15 | — | |
| PlanTFType=Learning2026.01 | 84.27 | 76.95 | — | |
| PlanTFType=Learning2026.01 | 84.27 | 76.95 | — | |
| PlanTFType=Learning-based2025.09 | 84.27 | 76.95 | — | |
| PlanTFType=Learning-based2026.02 | 84.27 | 76.95 | — | |
| GameFormerType=Learning, Post-processing=True2026.03 | 82.9 | 83.8 | — | |
| GameFormerType=Rule-based & Hybrid2025.09 | 79.94 | 79.78 | — | |
| GameFormerType=Rule-based & Hybrid2026.02 | 79.94 | 79.78 | — | |
| IDMType=Rule2026.01 | 75.6 | 77.33 | — | |
| IDMType=Rule2026.01 | 75.6 | 77.33 | — | |
| IDMType=Rule, Time (ms)=412025.12 | 75.6 | 77.33 | — | |
| IDMType=Rule-based & Hybrid2025.09 | 75.6 | 77.33 | — | |
| IDMType=Rule-based & Hybrid2026.02 | 75.6 | 77.33 | — | |
| IDMType=Rule, Post-processing=True2026.03 | 75.6 | 77.3 | — | |
| Raster-SoEType=Learning, Time (ms)=372025.12 | 72.05 | 71.47 | — | |
| RasterType=Learning, Time (ms)=372025.12 | 70.77 | 69.07 | — | |
| UrbanDriverType=Learning-based2026.02 | 68.57 | 64.11 | — | |
| PDM-Open-SoEType=Learning, Time (ms)=652025.12 | 54.09 | 56.73 | — | |
| PDM-OpenType=Learning, Time (ms)=652025.12 | 53.64 | 56.89 | — | |
| PDM-OpenType=Learning2026.01 | 53.53 | 54.24 | — | |
| PDM-OpenType=Learning2026.01 | 53.53 | 54.24 | — | |
| PDM-OpenType=Learning-based2025.09 | 53.53 | 54.24 | — | |
| PDM-OpenType=Learning-based, Pre-computed reference paths=true2026.02 | 53.53 | 54.24 | — | |
| GameFormerType=Learning2026.01 | 13.32 | 8.69 | — | |
| GameFormerType=Learning2026.01 | 13.32 | 8.69 | — | |
| GameFormerType=Learning-based, Refinement=None2025.09 | 13.32 | 8.69 | — | |
| GameFormer w/o refineType=Learning-based2026.02 | 13.32 | 8.69 | — | |
| DFPTraining Paradigm=Generative IL, Simulation Protocol=NR2026.06 | — | — | 90.33 | |
| DFPTraining Paradigm=Generative IL, Simulation Protocol=R2026.06 | — | — | 79.97 | |
| DFP-FMTraining Paradigm=Generative IL, Simulation Protocol=NR2026.06 | — | — | 92.68 | |
| DFP-FMTraining Paradigm=Generative IL, Simulation Protocol=R2026.06 | — | — | 81.3 | |
| Diffusion PlannerType=Rule-based & Hybrid, Refinement=true2025.11 | — | — | 94.24 | |
| Diffusion PlannerType=Learning-based2025.11 | — | — | 89.64 | |
| Diffusion PlannerTraining Paradigm=Generative IL, Simulation Protocol=NR2026.06 | — | — | 89.87 | |
| Diffusion PlannerTraining Paradigm=Generative IL, Simulation Protocol=R2026.06 | — | — | 82.8 | |
| ExpertTraining Paradigm=Human log replay, Simulation Protocol=NR2026.06 | — | — | 93.53 | |
| ExpertTraining Paradigm=Human log replay, Simulation Protocol=R2026.06 | — | — | 80.32 | |
| Flow PlannerTraining Paradigm=Generative IL, Simulation Protocol=NR2026.06 | — | — | 90.43 | |
| Flow PlannerTraining Paradigm=Generative IL, Simulation Protocol=R2026.06 | — | — | 83.31 | |
| GC-PGPType=Learning-based2025.11 | — | — | 59.51 | |
| IDMType=Rule-based & Hybrid2025.11 | — | — | 75.6 | |
| LAPType=Rule-based & Hybrid, Refinement=true, Solver order (o)=1, Denoising steps (s)=12025.11 | — | — | 94.7 | |
| LAPType=Rule-based & Hybrid, Refinement=true, Solver order (o)=1, Denoising steps (s)=22025.11 | — | — | 95.03 | |
| LAPType=Learning-based, Solver order (o)=1, Denoising steps (s)=12025.11 | — | — | 89.15 | |
| LAPType=Learning-based, Solver order (o)=1, Denoising steps (s)=22025.11 | — | — | 89.37 | |
| Log-ReplayType=Expert2025.11 | — | — | 93.53 | |
| PDM-ClosedType=Rule-based & Hybrid2025.11 | — | — | 92.84 | |
| PDM-HybridType=Rule-based & Hybrid2025.11 | — | — | 92.77 | |
| PDM-OpenType=Learning-based, Pre-searched reference lines=true2025.11 | — | — | 53.53 | |
| PDM-OpenTraining Paradigm=Imitation learning, Simulation Protocol=NR2026.06 | — | — | 53.53 | |
| PDM-OpenTraining Paradigm=Imitation learning, Simulation Protocol=R2026.06 | — | — | 54.24 | |
| Plan-R1Training Paradigm=IL + RL alignment, Simulation Protocol=NR2026.06 | — | — | 88.98 |