paper-with-me

홈 › Papers

DriveMA: Driving Vision-Language-Action Models with verifiable Meta-Actions

2026-05-29 · Weicheng Zheng, Yixin Huang, Qiao Sun, Derun Li, Hang Zhao arxiv

Driving Vision-Language-Action Models (Driving VLAs) aim to use language to improve end-to-end planning, but the language-action gap limits this promise. We propose DriveMA, a Driving VLA framework built on verifiable meta-actions, which summarize future ego motion into compact language-domain intentions and can be constructed from expert trajectories with a trajectory-grounded annotation pipeline and can be verified against generated trajectories through rule-based projection. DriveMA exploits this verifiability with action-centric supervised training and a data-efficient turn-level credit assignment reinforcement learning framework, explicitly aligning high-level decisions with low-level trajectory planning through dense rewards and precise credit assignment. DriveMA sets a new state of the art on the Waymo Open Dataset Vision-based E2E Driving, achieving a Rater Feedback Score of 8.060 with a 2B model and further improving it to 8.079 with a 4B model; it also obtains competitive closed-loop planning performance on NAVSIM. These results show that even a simple meta-action interface can achieve state-of-the-art planning when made verifiable and optimized for language-action alignment. Code, data, and models are available at https://tsinghua-mars-lab.github.io/DriveMA.

📄 PDF Abstract BibTeX arXiv:2605.31271

Code (0)

등록된 구현이 없습니다.

Tasks

Reinforcement LearningTrajectory Planning

Similar Papers 제목 키워드 기반

DriveMA: Rethinking Language Interfaces in Driving VLAs with One-Step Meta-Actions

2026-05-20 · Weicheng Zheng, Yixin Huang, Qiao Sun, Derun Li 외 arxiv

Driving Vision-Language-Action Models (Driving VLAs) commonly introduce natural-language reasoning as an intermediate interface for end-to-end planning, but reasoning-centric interfaces face three practical bottlenecks: …

Reinforcement Learning

DriveMamba: Task-Centric Scalable State Space Model for Efficient End-to-End Autonomous Driving

2026-02-09 · Haisheng Su, Wei Wu, Feixiang Song, Junjie Zhang 외 arxiv

Recent advances towards End-to-End Autonomous Driving (E2E-AD) have been often devoted on integrating modular designs into a unified framework for joint optimization e.g. UniAD, which follow a sequential paradigm (i.e., …

Autonomous Driving

A knowledge-augmented dataset of high-risk driving scenarios with LLM annotations for autonomous driving

2026-07-08 · Heye Huang, Jingguang Li, Zhiyuan Zhou, Paul Liang 외 arxiv

Safe autonomous driving requires both rapid responses to common high-risk events and deeper reasoning over rare, extreme long-tail scenarios in traffic safety. These scenarios are severely under-represented in naturalist…

Autonomous Driving

Deferred Exposure of Future Trajectories for Verifiable Reasoning in Autonomous Driving VLMs

2026-08-03 · Zixuan Huang, Yang Zhou, Kaixuan Wang, Guli Zhang 외 arxiv

Recent Vision-Language-Action (VLA) models for autonomous driving (AD) increasingly utilize chain-of-thought (CoT) supervision to enhance the reasoning capabilities of their Vision-Language Model (VLM) components, yet ex…

Autonomous Driving

VeriDrive: Verifiable Counterfactual Supervision for Cost-Efficient Vision-Language Planning

2026-06-05 · Zikai Zhang, Hubert P. H. Shum, Toby P. Breckon arxiv

Vision-language driving models increasingly use reasoning supervision to bridge perception, prediction, and planning, but existing driving rationales are often free-form and expensive to generate with frontier models. We…