Research map / Vision-language and multimodal / Robot action and navigation
Drive My Way: Preference Alignment of Vision-Language-Action Model for Personalized Driving
Zehao Wang, Huaide Jiang, Shuaiwu Dong, Yuping Wang, Hang Qiu, Jiachen Li
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 |
|---|---|---|---|
| fine-tune with GRPO and safety reward (fixed weights) (ablation) | SimLingo · Bench2Drive (neutral) | Driving Score: 78.15 -> 81.96 | Improved |
| add style-aware reward weights instead of fixed weights (ablation) | DMW · Bench2Drive (aggressive vs conservative styles) | style separation (efficiency, comfort): Efficiency 253.10 -> 281.56 (aggressive); comfort 21.87 -> 34.62 (conservative) | Improved |
| add style-aware reward weights instead of fixed weights (ablation) | DMW · Bench2Drive (aggressive) | Driving Score: 82.19 -> 79.50 | Hurt |
| add style-aware reward weights instead of fixed weights (ablation) | DMW · Bench2Drive (conservative) | Driving Score: 81.48 -> 82.72 | Improved |
| use DMW instead of SimLingo under aggressive instruction | SimLingo · Bench2Drive (aggressive) | Efficiency: 247.60 -> 281.56 | Improved |
| use DMW user embeddings instead of MORL-PD preference vector | long-term alignment · unseen drivers D3, D4 | Alignment Score: 0.25 -> 0.83 (D3); 0.33 -> 0.83 (D4) | Improved |
| add masked adaptive average pooling to preference encoder (ablation) | DMW long-term preference encoder · driver D3 alignment | Alignment Score: 0.25 -> 0.83 | Improved |
| add masked adaptive average pooling to preference encoder (ablation) | DMW long-term preference encoder · driver D1 ratings | Average Rating: 6.1 -> 8.7 | 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.
- SimLingo — SimLingo: Vision-Only Closed-Loop Autonomous Driving with Language-Action Alignment (CVPR 2025) · better on 5 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
- Less Language, More Latents: Annotation-Efficient VLAs for Driving (NeurIPS 2026 · similarity 0.89)
- LaMPilot: An Open Benchmark Dataset for Autonomous Driving with Language Model Programs (CVPR 2024 · similarity 0.88)
- AutoVLA: A Vision-Language-Action Model for End-to-End Autonomous Driving with Adaptive Reasoning and Reinforcement Fine-Tuning (NeurIPS 2025 · similarity 0.88)
- Distill to Think, Foresee to Act: Cognitive-Physical Reinforcement Learning for Autonomous Driving (NeurIPS 2026 · similarity 0.86)
- Driving Everywhere with Large Language Model Policy Adaptation (CVPR 2024 · similarity 0.86)
- DriveDPO: Policy Learning via Safety DPO For End-to-End Autonomous Driving (NeurIPS 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.
