Research map / Label-efficient and robust learning / Recognition and long-tail
UniTeD: Unified Temporal Diffusion for Joint Perception and Planning in Autonomous Driving
Bo Zhao, Xinting Zhao, Naifan Li, Erkang Cheng, 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 |
|---|---|---|---|
| unify perception and planning in one decoder (diffusion planner) (ablation) | UniTeD · NAVSIM v1 | PDMS: 86.7 -> 89.5 | Improved |
| use diffusion planning instead of regression (separate) (ablation) | UniTeD · NAVSIM v1 | PDMS: 84.1 -> 86.7 | Improved |
| use diffusion planning instead of regression (unified) (ablation) | UniTeD · NAVSIM v1 | PDMS: 88.5 -> 89.5 | Improved |
| also diffuse perception queries, not planning only (ablation) | UniTeD · NAVSIM v1 | PDMS: 89.5 -> 90.2 | Improved |
| add memory queue (regression) (ablation) | UniTeD · NAVSIM v1 | PDMS: 88.2 -> 88.5 | Improved |
| add memory queue (diffusion) (ablation) | UniTeD · NAVSIM v1 | PDMS: 89.0 -> 89.4 | Improved |
| add Temporal Transition Module (ablation) | UniTeD · NAVSIM v1 | PDMS: 89.4 -> 90.2 | Improved |
| add Anchor Refresh Strategy (ablation) | UniTeD · NAVSIM v1 | PDMS: 88.2 -> 90.2 | Improved |
| raise ARS confidence threshold from 0.3 to 0.5 (ablation) | UniTeD · NAVSIM v1 | PDMS: 90.23 -> 89.89 | Hurt |
| use 5 denoising steps instead of 2 (ablation) | UniTeD · NAVSIM v1 | PDMS: 90.23 -> 89.95 | Hurt |
| use 1 denoising step instead of 2 (ablation) | UniTeD · NAVSIM v1 | PDMS: 90.23 -> 89.57 | Hurt |
| compare with DiffRefiner (TF++ 500K data) on Bench2Drive success rate | UniTeD · Bench2Drive | Success rate: DS 87.1 -> 87.3; SR 71.4 -> 70.0 | Mixed |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- HiP-AD — HiP-AD: Hierarchical and Multi-Granularity Planning with Deformable Attention for Autonomous Driving in a Single Decoder (ICCV 2025) · better on 6 of 6 shared results
- DriveSuprim — DriveSuprim: Towards Precise Trajectory Selection for End-to-End Planning (AAAI 2026) · 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
- ChainFlow-VLA: Causal Flow Planning with Vision-Language Models (NeurIPS 2026 · similarity 0.88)
- $AutoDrive\text{-}P^3$: Unified Chain of Perception–Prediction–Planning Thought via Reinforcement Fine-Tuning (ICLR 2026 · 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)
- DiffusionDrive: Truncated Diffusion Model for End-to-End Autonomous Driving (CVPR 2025 · similarity 0.88)
- Diffusion-ES: Gradient-free Planning with Diffusion for Autonomous and Instruction-guided Driving (CVPR 2024 · similarity 0.88)
- Don't Shake the Wheel: Momentum-Aware Planning in End-to-End Autonomous Driving (CVPR 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.
