Research map / Label-efficient and robust learning / Recognition and long-tail
End-to-End Autonomous Driving Without Costly Modularization and 3D Manual Annotation
Mingzhe Guo, Zhipeng Zhang, Yuan He, Ke Wang, Liping Jing, Haibin Ling
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 |
|---|---|---|---|
| add spatial objectness loss (ablation) | UAD · nuScenes (NoAvg) | avg L2 / collision: 3.18m -> 1.00m L2; 2.43% -> 0.71% collision | Improved |
| add circular update in dreaming decoder (ablation) | UAD · nuScenes (NoAvg) | avg L2: 1.82 -> 1.12 | Improved |
| add dreaming loss (ablation) | UAD · nuScenes (NoAvg) | avg L2: 1.82 -> 1.04 | Improved |
| add directional augmentation (ablation) | UAD · nuScenes (NoAvg) | avg L2: 0.97 -> 0.92 | Improved |
| add directional consistency (ablation) | UAD · nuScenes (NoAvg) | avg collision: 0.32 -> 0.19 | Improved |
| add angular design (partition and angular queries) (ablation) | UAD · nuScenes (NoAvg) | avg L2 / collision: 1.37 -> 0.90 L2; 1.37 -> 0.19 collision | Improved |
| add 3D detection head (ablation) | UAD · nuScenes (NoAvg) | avg L2 / collision: 0.90 -> 0.93 L2; 0.19 -> 0.25 collision | No clear effect |
| use ImageNet pretraining for Res101 (ablation) | UAD · nuScenes (NoAvg) | avg L2: 0.95 -> 0.91 | Improved |
| use NuImages pretraining instead of none for Res101 (ablation) | UAD · nuScenes (NoAvg) | avg collision: 0.28 -> 0.19 | Improved |
| use GT 2D boxes instead of GroundingDINO pseudo ROIs (Res101) (ablation) | UAD · nuScenes (NoAvg) | avg L2: 0.90 -> 0.84 | Improved |
| use GT 2D boxes instead of pseudo ROIs (Res50) (ablation) | UAD · nuScenes (NoAvg) | avg L2: 0.99 -> 0.96 | Improved |
| decrease partition angle from 4 to 1 degree (ablation) | UAD · nuScenes (NoAvg) | avg L2 / collision: 0.90 -> 0.85 L2; 0.19 -> 0.32 collision, FPS 7.2 -> 5.0 | Mixed |
Closest papers by meaning
- Bench2Drive: Towards Multi-Ability Benchmarking of Closed-Loop End-To-End Autonomous Driving (NeurIPS 2024 · similarity 0.89)
- Navigation-Guided Sparse Scene Representation for End-to-End Autonomous Driving (ICLR 2025 · similarity 0.89)
- VR-Drive: Viewpoint-Robust End-to-End Driving with Feed-Forward 3D Gaussian Splatting (NeurIPS 2025 · similarity 0.88)
- DriveTransformer: Unified Transformer for Scalable End-to-End Autonomous Driving (ICLR 2025 · similarity 0.88)
- Prioritizing Perception-Guided Self-Supervision: A New Paradigm for Causal Modeling in End-to-End Autonomous Driving (NeurIPS 2025 · similarity 0.88)
- Enhancing End-to-End Autonomous Driving with Latent World Model (ICLR 2025 · similarity 0.88)
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.
