Open-loop planning on nuScenes
0.18L2 Error (Avg)OWMDrive
Evaluation Results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| OWMDriveInput=C, Scene Representation=Det & Map & Motion2026.06 | 0.18 | 0.12 | 0.18 | 0.25 | 1 | 4 | 0.11 | 5 | — | |
| EvoDriveVLACategory=Distillation-Based, Evaluation Protocol=ST-P32026.03 | 0.26 | 0.12 | 0.24 | 0.43 | 2 | 5 | 0.12 | 0.06 | — | |
| DiMACategory=Distillation-Based, Evaluation Protocol=ST-P32026.03 | 0.27 | 0.12 | 0.25 | 0.44 | 4 | 6 | 0.15 | 0.08 | — | |
| SOLVE-VLMEgo=true, Model Type=Text-Based2025.12 | 0.28 | 0.13 | 0.25 | 0.47 | 0 | 16 | 43 | 20 | — | |
| EMMA+Ego=true, Model Type=Text-Based2025.12 | 0.29 | 0.13 | 0.27 | 0.48 | — | — | — | — | — | |
| ImpromptuVLAEgo=true, Model Type=Text-Based2025.12 | 0.3 | 0.13 | 0.27 | 0.53 | — | — | — | — | — | |
| ColaVLAEgo=true, Model Type=Action-Based2025.12 | 0.3 | 0.14 | 0.27 | 0.5 | 4 | 17 | 47 | 23 | — | |
| DriveVLM-DualEgo=true, Model Type=Text-Based2025.12 | 0.31 | 0.15 | 0.29 | 0.48 | — | — | — | — | — | |
| SOLVE-E2EEgo=true, Model Type=Action-Based2025.12 | 0.31 | 0.14 | 0.28 | 0.5 | 4 | 17 | 68 | 30 | — | |
| DriveVLM-DualEgo Status=Vector, Finetuning: Understanding=true, Finetuning: Decision=false, Finetuning: Planning=true, Evaluation Protocol=ST-P32026.03 | 0.31 | 0.15 | 0.29 | 0.48 | 5 | 8 | 17 | 10 | — | |
| DynFlowDrive⋄ (SSR)Pub.=-, BEV=✓, Backbone=ResNet-50, Ego status=true2026.03 | 0.31 | 0.13 | 0.27 | 0.52 | 6 | 9 | 0.18 | 11 | — | |
| EMMAEgo=true, Model Type=Text-Based2025.12 | 0.32 | 0.14 | 0.29 | 0.54 | — | — | — | — | — | |
| EMMA†Ego Status=Text, Finetuning: Understanding=true, Finetuning: Decision=false, Finetuning: Planning=true, Evaluation Protocol=ST-P32026.03 | 0.32 | 0.14 | 0.29 | 0.54 | — | — | — | — | — | |
| SpaceDriveEgo Status=Vector, Finetuning: Understanding=true, Finetuning: Decision=false, Finetuning: Planning=true, Evaluation Protocol=ST-P32026.03 | 0.32 | 0.15 | 0.29 | 0.51 | 4 | 18 | 49 | 23 | — | |
| AutoMoTEgo Status=Vector, Finetuning: Understanding=false, Finetuning: Decision=true, Finetuning: Planning=true, Evaluation Protocol=ST-P32026.03 | 0.32 | 0.14 | 0.29 | 0.54 | 1 | 6 | 15 | 7 | — | |
| OmniDriveEgo=true, Model Type=Text-Based2025.12 | 0.33 | 0.14 | 0.29 | — | 0 | 13 | 78 | 30 | — | |
| OmniDriveCategory=LLM-Based, Evaluation Protocol=ST-P32026.03 | 0.33 | 0.14 | 0.29 | 0.55 | 0 | 13 | 0.78 | 0.3 | — | |
| OpenDriveVLACategory=LLM-Based, Evaluation Protocol=ST-P32026.03 | 0.33 | 0.14 | 0.3 | 0.55 | 2 | 7 | 0.22 | 0.1 | — | |
| DriveTransformer-LargeEgo Status=Vector, Finetuning: Understanding=false, Finetuning: Decision=false, Finetuning: Planning=true, Evaluation Protocol=ST-P32026.03 | 0.33 | 0.16 | 0.3 | 0.55 | 1 | 6 | 15 | 7 | — | |
| RoboTron-DriveFinetuning: Understanding=true, Finetuning: Decision=false, Finetuning: Planning=true, Evaluation Protocol=ST-P32026.03 | 0.33 | 0.14 | 0.3 | 0.57 | 3 | 12 | 63 | 26 | — | |
| OpenDrive-VLAEgo Status=Text, Finetuning: Understanding=true, Finetuning: Decision=false, Finetuning: Planning=true, Evaluation Protocol=ST-P32026.03 | 0.33 | 0.15 | 0.31 | 0.55 | 1 | 8 | 21 | 10 | — | |
| OmniDriveEgo Status=Vector, Finetuning: Understanding=true, Finetuning: Decision=false, Finetuning: Planning=true, Evaluation Protocol=ST-P32026.03 | 0.33 | 0.14 | 0.29 | 0.55 | 0 | 13 | 78 | 30 | — | |
| ORIONCategory=LLM-Based, Evaluation Protocol=ST-P32026.03 | 0.34 | 0.17 | 0.31 | 0.55 | 5 | 25 | 0.8 | 0.37 | — | |
| ORION(Chat-B2D)Finetuning: Understanding=true, Finetuning: Decision=false, Finetuning: Planning=true, Evaluation Protocol=ST-P32026.03 | 0.34 | 0.17 | 0.31 | 0.55 | 5 | 25 | 80 | 37 | — | |
| ORION2026.03 | 0.34 | 0.17 | 0.31 | 0.55 | 5 | 25 | 0.8 | 37 | — | |
| AD-MLPEgo=true, Model Type=Action-Based2025.12 | 0.35 | 0.15 | 0.32 | 0.59 | 0 | 27 | 85 | 37 | — | |
| BEV-Planner++Ego=true, Model Type=Action-Based2025.12 | 0.35 | 0.16 | 0.32 | 0.57 | 0 | 29 | 73 | 34 | — | |
| BEV-PlannerCategory=Traditional, Evaluation Protocol=ST-P32026.03 | 0.35 | 0.16 | 0.32 | 0.57 | 0 | 29 | 0.73 | 0.34 | — | |
| Ego-MLPEgo Status=Vector, Finetuning: Understanding=false, Finetuning: Decision=false, Finetuning: Planning=true, Evaluation Protocol=ST-P32026.03 | 0.35 | 0.15 | 0.32 | 0.59 | 0 | 27 | 85 | 37 | — | |
| DynFlowDrive (SSR)Pub.=-, BEV=✓, Backbone=ResNet-502026.03 | 0.35 | 0.16 | 0.32 | 0.57 | 7 | 9 | 0.21 | 14 | — | |
| BEVPlaner++Train Data=nuScenes, Use perception annotations=false2026.05 | 0.35 | 0.16 | 0.32 | 0.57 | 0 | 0.29 | 0.73 | 0.34 | — | |
| HEATTrain Data=nuScenes, NAVSIM, Waymo, Use perception annotations=false2026.05 | 0.35 | 0.15 | 0.32 | 0.58 | 0.1 | 0.13 | 0.28 | 0.17 | — | |
| OpenREADEgo Status=Vector, Finetuning: Understanding=true, Finetuning: Decision=false, Finetuning: Planning=true, Evaluation Protocol=ST-P32026.03 | 0.36 | 0.17 | 0.34 | 0.56 | 4 | 8 | 22 | 11 | — | |
| VAD-BaseEgo=true, Model Type=Action-Based2025.12 | 0.37 | 0.17 | 0.34 | 0.6 | 4 | 27 | 67 | 33 | — | |
| VADCategory=Traditional, Evaluation Protocol=ST-P32026.03 | 0.37 | 0.17 | 0.34 | 0.6 | 4 | 27 | 0.67 | 0.33 | — | |
| VADEgo Status=Vector, Finetuning: Understanding=false, Finetuning: Decision=false, Finetuning: Planning=true, Evaluation Protocol=ST-P32026.03 | 0.37 | 0.17 | 0.34 | 0.6 | 7 | 10 | 24 | 14 | — | |
| VAD-BaseTrain Data=nuScenes, Use perception annotations=true2026.05 | 0.37 | 0.17 | 0.34 | 0.6 | 0.07 | 0.1 | 0.24 | 0.14 | — | |
| LaST-VLA*VLM Model=InternVL3-8B, Ego status=True2026.03 | 0.38 | 0.17 | 0.33 | 0.64 | 0 | 11 | 0.42 | 18 | — | |
| BEVPlaner++Train Data=nuScenes, NAVSIM, Waymo, Use perception annotations=false2026.05 | 0.38 | 0.17 | 0.34 | 0.62 | 0 | 0.33 | 1 | 0.44 | — | |
| SSR*Pub.=ICLR 2025, BEV=✓, Backbone=ResNet-50, Implementation=re-implemented from SSR by official public codes2026.03 | 0.39 | 0.18 | 0.35 | 0.63 | 8 | 12 | 0.24 | 15 | — | |
| DriveVLMEgo=true, Model Type=Text-Based2025.12 | 0.4 | 0.18 | 0.34 | 0.68 | — | — | — | — | — | |
| DriveVLMCategory=LLM-Based, Evaluation Protocol=ST-P32026.03 | 0.4 | 0.18 | 0.34 | 0.68 | 10 | 22 | 0.45 | 0.27 | — | |
| AutoVLAEgo Status=Text, Finetuning: Understanding=true, Finetuning: Decision=false, Finetuning: Planning=true, Evaluation Protocol=ST-P32026.03 | 0.4 | 0.21 | 0.38 | 0.6 | 13 | 18 | 28 | 20 | — | |
| DriveVLMRequires VLM @ Inference=true, Ego Status=true2025.05 | 0.4 | 0.18 | 0.34 | 0.68 | — | — | — | — | 2.43 | |
| VAD-TinyTrain Data=nuScenes, Use perception annotations=true2026.05 | 0.41 | 0.2 | 0.38 | 0.65 | 0.1 | 0.12 | 0.27 | 0.16 | — | |
| LaST-VLA*VLM Model=InternVL3-2B, Ego status=True2026.03 | 0.42 | 0.19 | 0.35 | 0.71 | 3 | 16 | 0.47 | 22 | — | |
| GPT-DriverCategory=LLM-Based, Evaluation Protocol=ST-P32026.03 | 0.44 | 0.2 | 0.4 | 0.7 | 4 | 12 | 0.36 | 0.17 | — | |
| FSDrive*VLM Model=Qwen2-VL-2B, Ego status=True2026.03 | 0.45 | 0.18 | 0.39 | 0.77 | 0 | 6 | 0.42 | 16 | — | |
| UniADEgo=true, Model Type=Action-Based2025.12 | 0.46 | 0.2 | 0.42 | 0.75 | 2 | 25 | 84 | 37 | — | |
| UniAD*VLM Model=-, Ego status=True2026.03 | 0.46 | 0.2 | 0.42 | 0.75 | 2 | 25 | 0.84 | 37 | — | |
| BEV-PlannerPub.=CVPR 2024, BEV=✓, Backbone=ResNet-502026.03 | 0.46 | 0.28 | 0.42 | 0.68 | 4 | 37 | 1.07 | 49 | — | |
| BEV-PlannerInput=C, Scene Representation=None2026.06 | 0.46 | 0.28 | 0.42 | 0.68 | 4 | 37 | 1.07 | 49 | — | |
| Drive-OccWorldPub.=AAAI 2025, BEV=✗, Backbone=ResNet-502026.03 | 0.47 | 0.25 | 0.44 | 0.72 | 3 | 8 | 0.22 | 11 | — | |
| PARA-DrivePub.=CVPR 2024, BEV=✓, Backbone=ResNet-502026.03 | 0.48 | 0.25 | 0.46 | 0.74 | 14 | 23 | 0.39 | 25 | — | |
| PARA-DriveTrain Data=nuScenes, Use perception annotations=true2026.05 | 0.48 | 0.25 | 0.46 | 0.74 | 0.14 | 0.23 | 0.39 | 0.25 | — | |
| PARA-DriveInput=C, Scene Representation=Det & Map & Motion2026.06 | 0.48 | 0.25 | 0.46 | 0.74 | 14 | 23 | 0.39 | 25 | — | |
| World4DriveTrain Data=nuScenes, Use perception annotations=false2026.05 | 0.5 | 0.23 | 0.47 | 0.81 | 0.02 | 0.12 | 0.33 | 0.16 | — | |
| EvoDriveVLACategory=Distillation-Based, Evaluation Protocol=UniAD2026.03 | 0.52 | 0.16 | 0.44 | 0.96 | 2 | 2 | 0.33 | 0.12 | — | |
| GenADPub.=ECCV 2024, BEV=✓, Backbone=ResNet-502026.03 | 0.52 | 0.28 | 0.49 | 0.78 | 8 | 14 | 0.34 | 19 | — | |
| GenADInput=C, Scene Representation=Det & Map & Motion2026.06 | 0.52 | 0.28 | 0.58 | 0.96 | 8 | 14 | 0.34 | 19 | — | |
| CausalVADFPS=5.42026.03 | 0.54 | 0.27 | 0.52 | 0.82 | 2 | 9 | 0.22 | 11 | — | |
| HEATTrain Data=nuScenes, Use perception annotations=false2026.05 | 0.54 | 0.24 | 0.53 | 0.86 | 0.08 | 0.16 | 0.44 | 0.23 | — | |
| BEV-PlannerEgo=false, Model Type=Action-Based2025.12 | 0.55 | 0.3 | 0.52 | 0.83 | 10 | 37 | 130 | 59 | — | |
| DiffusionDriveCategory=Traditional, Evaluation Protocol=ST-P32026.03 | 0.57 | 0.27 | 0.54 | 0.9 | 3 | 5 | 0.16 | 0.08 | — | |
| DistillDriveCategory=Distillation-Based, Evaluation Protocol=ST-P32026.03 | 0.57 | 0.28 | 0.54 | 0.83 | 0 | 3 | 0.17 | 0.06 | — | |
| DiMACategory=Distillation-Based, Evaluation Protocol=UniAD2026.03 | 0.57 | 0.18 | 0.48 | 1.01 | 0 | 5 | 0.16 | 0.07 | — | |
| DiffusionDrivePub.=CVPR 2025, BEV=✗, Backbone=ResNet-502026.03 | 0.57 | 0.27 | 0.54 | 0.9 | 3 | 5 | 0.16 | 8 | — | |
| DynFlowDrive(LAW)Pub.=-, BEV=✗, Backbone=Swin-Tiny2026.03 | 0.57 | 0.24 | 0.54 | 0.92 | 11 | 12 | 0.48 | 22 | — | |
| DiffusionDriveTrain Data=nuScenes, Use perception annotations=true2026.05 | 0.57 | 0.27 | 0.54 | 0.9 | 0.03 | 0.05 | 0.16 | 0.08 | — | |
| BEVPlanerTrain Data=nuScenes, Use perception annotations=false2026.05 | 0.57 | 0.27 | 0.54 | 0.9 | 0.04 | 0.35 | 1.8 | 0.73 | — | |
| DiffusionDriveInput=C, Scene Representation=Det & Map & Motion2026.06 | 0.57 | 0.27 | 0.54 | 0.9 | 3 | 5 | 0.16 | 8 | — | |
| EchoVLAPerception Modality=Vision + Audio2026.01 | 0.58 | 0.46 | 0.52 | 0.74 | 0 | 12 | 22 | 11 | — | |
| PPADFPS=2.62026.03 | 0.58 | 0.31 | 0.56 | 0.87 | 8 | 12 | 0.38 | 19 | — | |
| BridgeADFPS=3.92026.03 | 0.58 | 0.28 | 0.55 | 0.92 | 0 | 4 | 0.2 | 8 | — | |
| Senna2026.03 | 0.59 | 0.37 | 0.54 | 0.86 | 9 | 12 | 0.33 | 18 | — | |
| MomADPub.=CVPR 2025, BEV=✗, Backbone=ResNet-502026.03 | 0.6 | 0.31 | 0.57 | 0.91 | 1 | 5 | 0.22 | 9 | — | |
| SparseDriveFPS=6.12026.03 | 0.61 | 0.3 | 0.58 | 0.95 | 1 | 5 | 0.23 | 10 | — | |
| SparseDrivePub.=ICCV 2025, BEV=✗, Backbone=ResNet-502026.03 | 0.61 | 0.29 | 0.58 | 0.96 | 1 | 5 | 0.18 | 8 | — | |
| LAWPub.=ICLR 2025, BEV=✗, Backbone=Swin-Tiny2026.03 | 0.61 | 0.26 | 0.57 | 1.01 | 14 | 21 | 0.54 | 30 | — | |
| SparseDriveTrain Data=nuScenes, Use perception annotations=true2026.05 | 0.61 | 0.29 | 0.58 | 0.96 | 0.01 | 0.05 | 0.18 | 0.08 | — | |
| LAWTrain Data=nuScenes, Use perception annotations=false2026.05 | 0.61 | 0.26 | 0.57 | 1.01 | 0.14 | 0.21 | 0.54 | 0.3 | — | |
| SparseDriveInput=C, Scene Representation=Det & Map & Motion2026.06 | 0.61 | 0.29 | 0.58 | 0.96 | 5 | 18 | 0.34 | 18 | — | |
| VAD §FPS=3.12026.03 | 0.62 | 0.34 | 0.59 | 0.92 | 21 | 34 | 0.58 | 38 | — | |
| VeteranADFPS=4.32026.03 | 0.62 | 0.33 | 0.59 | 0.94 | 3 | 12 | 0.34 | 16 | — | |
| VERDI2025.05 | 0.65 | 0.36 | 0.62 | 0.96 | — | — | — | — | 4.5 | |
| OpenDriveVLA*VLM Model=Qwen2.5VL-3B, Ego status=True2026.03 | 0.67 | 0.19 | 0.58 | 1.24 | 2 | 18 | 0.7 | 30 | — | |
| OpenDriveVLACategory=LLM-Based, Evaluation Protocol=UniAD2026.03 | 0.67 | 0.19 | 0.58 | 1.24 | 2 | 18 | 0.7 | 0.3 | — | |
| UniADCategory=Traditional, Evaluation Protocol=ST-P32026.03 | 0.69 | 0.44 | 0.67 | 0.96 | 4 | 8 | 0.23 | 0.12 | — | |
| UniADEgo Status=Vector, Finetuning: Understanding=false, Finetuning: Decision=false, Finetuning: Planning=true, Evaluation Protocol=ST-P32026.03 | 0.69 | 0.44 | 0.67 | 0.96 | 4 | 8 | 23 | 12 | — | |
| UniADInput=C, Scene Representation=Det & Map & Occ & Motion2026.06 | 0.69 | 0.44 | 0.67 | 0.96 | 4 | 8 | 0.23 | 12 | — | |
| AutoVLA*VLM Model=Qwen2.5-VL-3B, Ego status=True2026.03 | 0.7 | 0.28 | 0.66 | 1.16 | 14 | 25 | 0.53 | 31 | — | |
| LTFTrain Data=nuScenes, Use perception annotations=false2026.05 | 0.7 | 0.45 | 0.66 | 0.99 | 0.08 | 0.08 | 0.23 | 0.13 | — | |
| VAD-BasePub.=ICCV 2023, BEV=✓, Backbone=ResNet-502026.03 | 0.72 | 0.41 | 0.7 | 1.05 | 7 | 17 | 0.41 | 22 | — | |
| VAD-Base2025.05 | 0.72 | 0.41 | 0.7 | 1.05 | — | — | — | — | 4.5 | |
| VADInput=C, Scene Representation=Det & Map & Motion2026.06 | 0.72 | 0.41 | 0.7 | 1.05 | 7 | 17 | 0.41 | 22 | — | |
| UniADFPS=1.82026.03 | 0.73 | 0.45 | 0.7 | 1.04 | 62 | 58 | 0.63 | 61 | — | |
| VLP-UniADPerception Modality=Vision2026.01 | 0.74 | 0.36 | 0.68 | 1.19 | 3 | 12 | 32 | 16 | — | |
| VAD-tiny §FPS=5.62026.03 | 0.74 | 0.44 | 0.72 | 1.06 | 31 | 48 | 0.52 | 44 | — | |
| UniADTrain Data=nuScenes, Use perception annotations=true2026.05 | 0.76 | 0.48 | 0.74 | 1.07 | 0.12 | 0.13 | 0.28 | 0.17 | — | |
| OccWorldCategory=LLM-Based, Evaluation Protocol=ST-P32026.03 | 0.77 | 0.39 | 0.73 | 1.18 | 11 | 19 | 0.67 | 0.32 | — |