Research map / Label-efficient and robust learning / Recognition and long-tail
VAD: Vectorized Scene Representation for Efficient Autonomous Driving
Bo Jiang, Shaoyu Chen, Qing Xu, Bencheng Liao, Jiajie Chen, Helong Zhou, Qian Zhang, Wenyu Liu, Chang Huang, Xinggang Wang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use vectorized scene representation and planning instead of UniAD's rasterized pipeline | VAD-Base · nuScenes val open-loop planning | average collision rate (%): 0.31 -> 0.22 | Improved |
| use vectorized scene representation and planning instead of UniAD's rasterized pipeline | VAD-Base · nuScenes val open-loop planning | average L2 error (m): 1.03 -> 0.72 | Improved |
| drop auxiliary tasks (tracking, occupancy) and post-processing with vectorized representation (VAD-Tiny) | VAD · nuScenes val open-loop planning | FPS: 1.8 -> 16.8 | Improved |
| use vectorized map instead of rasterized map representation (no map constraints) (ablation) | VAD · nuScenes val | average collision rate (%): 0.39 -> 0.26 | Improved |
| use vectorized map instead of rasterized map representation (no map constraints) (ablation) | VAD · nuScenes val | average L2 error (m): 0.74 -> 0.77 | Hurt |
| add ego-map interaction in planning (ablation) | VAD · nuScenes val | average L2 error (m): 0.86 -> 0.72 | Improved |
| add ego-agent interaction in planning (ablation) | VAD · nuScenes val | average collision rate (%): 0.26 -> 0.22 | Improved |
| add all three vectorized planning constraints (ablation) | VAD · nuScenes val | average collision rate (%): 0.28 -> 0.22 | Improved |
| add ego-boundary overstepping constraint only (ablation) | VAD · nuScenes val | average collision rate (%): 0.28 -> 0.24 | Improved |
| add ego-agent collision constraint only (ablation) | VAD · nuScenes val | average L2 error (m): 0.76 -> 0.77 | No clear effect |
| give ego status as input to planning | VAD-Base · nuScenes val open-loop planning | average L2 error (m): 0.72 -> 0.37 | Improved |
| use VAD-Base instead of UniAD on 1s collision rate | VAD-Base · nuScenes val open-loop planning | 1s collision rate (%): 0.05 (UniAD) vs 0.07 (VAD-Base) | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- ST-P3 — ST-P3: End-to-End Vision-Based Autonomous Driving via Spatial-Temporal Feature Learning (ECCV 2022) · better on 4 of 5 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- VADv2: End-to-End Autonomous Driving via Probabilistic Planning (ICLR 2026 · similarity 0.90)
- SGDrive: Scene-to-Goal Hierarchical World Cognition for Autonomous Driving (CVPR 2026 · similarity 0.87)
- Scenario Dreamer: Vectorized Latent Diffusion for Generating Driving Simulation Environments (CVPR 2025 · similarity 0.87)
- Future-Aware End-to-End Driving: Bidirectional Modeling of Trajectory Planning and Scene Evolution (NeurIPS 2025 · similarity 0.87)
- VR-Drive: Viewpoint-Robust End-to-End Driving with Feed-Forward 3D Gaussian Splatting (NeurIPS 2025 · similarity 0.86)
- Navigation-Guided Sparse Scene Representation for End-to-End Autonomous Driving (ICLR 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
