Research map / Vision-language and multimodal / Robot action and navigation
DriveFine: Refining-Augmented Masked Diffusion VLA for Accurate and Robust Driving
Chenxu Dang, Sining Ang, Yongkang Li, Haochen Tian, Jie Wang, Guang Li, Hangjun Ye, Jie Ma, Long Chen, Yan 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 |
|---|---|---|---|
| add GRPO reinforcement training to SFT (ablation) | DriveFine (LLaDA) · NAVSIM-v1 navtest | PDMS: 86.7 -> 89.6 | Improved |
| add refinement expert with offline RFT (ablation) | DriveFine · NAVSIM-v1 navtest | PDMS: 89.6 -> 90.3 | Improved |
| add online RFT on top of offline RFT (ablation) | DriveFine · NAVSIM-v1 navtest | PDMS: 90.3 -> 90.8 | Improved |
| add 1 refinement block (250M params) versus none (ablation) | DriveFine · NAVSIM-v1 navtest | PDMS: 89.6 -> 90.0 | Improved |
| increase refinement blocks from 4 to 8 (ablation) | DriveFine · NAVSIM-v1 navtest | PDMS: 90.8 -> 90.7 | No clear effect |
| increase GRPO group size from 2 to 8 (ablation) | DriveFine · NAVSIM-v1 navtest | PDMS: 88.4 -> 89.6 | Improved |
| increase GRPO group size from 8 to 10 (ablation) | DriveFine · NAVSIM-v1 navtest | PDMS: 89.6 -> 89.6 | No clear effect |
| add PDMS-oriented RFT to SFT model (ablation) | DriveFine · NAVSIM-v1 navtest | PDMS: 86.7 -> 90.5 | Improved |
| add PDMS-oriented RFT to SFT model (ablation) | DriveFine · NAVSIM-v1 navtest | EPDMS: 86.8 -> 89.1 | Improved |
| add PDMS-oriented RFT to SFT model (ablation) | DriveFine · NAVSIM-v1 navtest | LK: 97.9 -> 97.7 | No clear effect |
| add PDMS-oriented RFT to SFT model (ablation) | DriveFine · NAVSIM-v1 navtest | EC: 83.3 -> 83.8 | No clear effect |
| add score-based reinforcement fine-tuning with an extra scorer | DriveFine · NAVSIM-v1 navtest | PDMS: 90.7 -> 91.8 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- ReCogDrive — ReCogDrive: A Reinforced Cognitive Framework for End-to-End Autonomous Driving (ICLR 2026) · better on 4 of 4 shared results
- DriveVLA-W0 — DriveVLA-W0: World Models Amplify Data Scaling Law in Autonomous Driving (ICLR 2026) · better on 3 of 3 shared results
- AutoVLA — AutoVLA: A Vision-Language-Action Model for End-to-End Autonomous Driving with Adaptive Reasoning and Reinforcement Fine-Tuning (NeurIPS 2025) · better on 2 of 2 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
- ReflectDrive-2: Reinforcement-Learning-Aligned Self-Editing for Discrete Diffusion Driving (NeurIPS 2026 · similarity 0.90)
- DiffusionDrive: Truncated Diffusion Model for End-to-End Autonomous Driving (CVPR 2025 · similarity 0.90)
- ChainFlow-VLA: Causal Flow Planning with Vision-Language Models (NeurIPS 2026 · similarity 0.89)
- FIVE-VLA: Fast and EffectIVE Closed-Loop Autonomous Driving with Recurrent Action Memory (NeurIPS 2026 · similarity 0.89)
- Diffusion-ES: Gradient-free Planning with Diffusion for Autonomous and Instruction-guided Driving (CVPR 2024 · similarity 0.89)
- AutoVLA: A Vision-Language-Action Model for End-to-End Autonomous Driving with Adaptive Reasoning and Reinforcement Fine-Tuning (NeurIPS 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.
