Research map / Label-efficient and robust learning / Recognition and long-tail
Plug-and-Play Traffic Element Awareness for End-to-End Autonomous Driving
Zongzheng Zhang, Jijun Wang, Saining Zhang, Wang Shuo, Yiru Wang, Hai Yang, Yang Chen, Yuwen Heng, HAO SUN, Jiang anqing, HAO ZHAO
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 auxiliary 3D traffic element supervision | VAD · nuScenes | L2 avg (m): 0.72 -> 0.63 | Improved |
| add language-encoded ego-centric topology on top of TE | Orion · nuScenes | L2 avg (m): 0.29 -> 0.26 | Improved |
| add TE supervision and conditioning | LTF · NAVSIM-v2 navhard | EPDMS: 25.1 -> 28.9 | Improved |
| add TE supervision and conditioning | VAD · Bench2Drive | Driving Score: 42.3 -> 56.4 | Improved |
| add TE supervision and topology | VAD · Bench2Drive | Efficiency: 157.94 -> 125.64 | Hurt |
| use explicit 3D TE instead of 2D front-view TE (ablation) | LTF · NAVSIM-v2 | EPDMS: 28.1 -> 28.9 | Improved |
| use targeted TE depth instead of full front-view depth (ablation) | LTF · NAVSIM-v2 | EPDMS: 27.7 -> 28.9 | Improved |
| estimate TE 3D centers with monocular depth fused with LiDAR instead of LiDAR clustering (ablation) | LTF · NAVSIM-v2 | EPDMS: 27.9 -> 28.9 | Improved |
| use UniDepthV2 instead of Depth Anything 3 (ablation) | LTF · NAVSIM-v2 | EPDMS: 27.8 -> 28.9 | Improved |
| model traffic signs in addition to traffic lights (ablation) | LTF · NAVSIM-v2 | EPDMS: 26.7 -> 28.9 | Improved |
| use independent TE head instead of BEV semantic class with cross-entropy (ablation) | LTF · NAVSIM-v2 | EPDMS: 24.3 -> 26.1 | Improved |
| use focal loss instead of cross-entropy under independent head (ablation) | LTF · NAVSIM-v2 | EPDMS: 25.4 -> 26.1 | Improved |
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.
- DriveTransformer-Large — DriveTransformer: Unified Transformer for Scalable End-to-End Autonomous Driving (ICLR 2025) · better on 1 of 1 shared result
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
- Bench2Drive: Towards Multi-Ability Benchmarking of Closed-Loop End-To-End Autonomous Driving (NeurIPS 2024 · similarity 0.88)
- Dual-View Interaction-Aware Lane Change Prediction for Autonomous Driving (AAAI 2025 · similarity 0.88)
- DriveHierarchy: A Benchmark for Diagnosing VLM Driving Capabilities from Open-Loop Understanding to Closed-Loop Execution (NeurIPS 2026 · similarity 0.87)
- Prioritizing Perception-Guided Self-Supervision: A New Paradigm for Causal Modeling in End-to-End Autonomous Driving (NeurIPS 2025 · similarity 0.87)
- DriveCombo: Benchmarking Compositional Traffic Rule Reasoning in Autonomous Driving (CVPR 2026 · similarity 0.87)
- Beyond Self-Play and Scale: A Behavior Benchmark for Generalization in Autonomous Driving (NeurIPS 2026 · 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.
