Autonomous Driving Planning on WOD-E2E (val)
8.2RFS ScoreMindVLA-U1 + RL
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| MindVLA-U1 + RLBackbone=Qwen3-VL-2B, AD-PT=×2026.05 | 8.2 | 1.01 | 2.28 | 0.51 | 1.03 | |
| Human Driver2026.05 | 8.13 | — | — | — | — | |
| Poutine-BaseBackbone=Qwen2.5-VL-3B, AD-PT=✓2026.05 | 8.12 | 1.27 | 2.94 | — | — | |
| MindVLA-U1 + Intent-CFGBackbone=Qwen3-VL-2B, AD-PT=×2026.05 | 7.92 | 0.86 | 2.13 | 0.47 | 1.07 | |
| MindVLA-U1 + MoTBackbone=Qwen3-VL-2B, AD-PT=×2026.05 | 7.92 | 0.89 | 2.11 | 0.49 | 1.05 | |
| RAP-DINOBackbone=DINOv3-H, AD-PT=✓2026.05 | 7.91 | 0.97 | 2.2 | — | — | |
| MindVLA-U1Backbone=Qwen3-VL-2B, AD-PT=×2026.05 | 7.83 | 0.92 | 2.14 | 0.5 | 1.05 | |
| UniADBackbone=ImageNet ViT, AD-PT=×2026.05 | 5.78 | 6.5 | 10.81 | — | — | |
| VADBackbone=ImageNet vision encoder, AD-PT=×2026.05 | 4.45 | 3.19 | 5.81 | — | — |