外观
论文arXiv 2512.1363610 个月前
MindDrive：双LoRA架构VLA模型实现自动驾驶在线强化学习
本文由 AI 根据公开内容整理，可能有误，请以原文为准：MindDrive: A Vision-Language-Action Model for Autonomous Driving via Online Reinforcement Learning（arXiv 2512.13636）
核心内容
针对自动驾驶VLA模型依赖模仿学习导致的分布偏移与因果混淆问题，研究团队提出MindDrive框架，首次证实在线强化学习在VLA自动驾驶中的有效性。该方法采用轻量级Qwen-0.5B作为基座，在Bench2Drive高难度基准测试中实现78.04的Driving Score和55.09%的成功率，显著优于传统纯模仿学习方案，为具身智能体在复杂动态环境下的策略优化提供了新路径。
技术上，MindDrive在单一LLM内集成两组独立LoRA参数：Decision Expert负责场景推理与离散化驾驶决策，Action Expert将语言指令动态映射为可行轨迹。系统通过将轨迹级奖励反馈至推理空间，使试错学习限定在有限离散语言决策集内，规避了连续动作空间探索效率低下的瓶颈。这种解耦设计既保留了大模型的语义理解能力，又实现了符合人类驾驶习惯的策略寻优与高效收敛。
关键看点
- 双LoRA解耦架构：单LLM内嵌Decision与Action两组LoRA，分别处理场景推理决策与语言到轨迹的动态映射
- 离散化RL探索：将连续动作空间转化为有限离散语言决策集，通过轨迹级奖励反馈实现高效试错学习
- 轻量化基准性能：基于Qwen-0.5B在Bench2Drive获78.04驾驶分与55.09%成功率，验证在线RL有效性
- 解决IL固有缺陷：针对性克服模仿学习的分布偏移与因果混淆，平衡复杂场景决策最优性与类人驾驶行为
工程与产业观察
MindDrive将VLA在线强化学习从连续控制降维至离散语言空间，为具身智能体提供了可迁移的策略优化范式。其双LoRA解耦设计与小参数量高性能表现，对机器人端侧部署及Sim-to-Real训练具有直接参考价值，证明了语言表征可作为连接高层认知与底层控制的高效强化学习接口。
英文摘要 (Abstract)
Current Vision-Language-Action (VLA) paradigms in autonomous driving primarily rely on Imitation Learning (IL), which introduces inherent challenges such as distribution shift and causal confusion. Online Reinforcement Learning offers a promising pathway to address these issues through trial-and-error learning. However, applying online reinforcement learning to VLA models in autonomous driving is hindered by inefficient exploration in continuous action spaces. To overcome this limitation, we propose MindDrive, a VLA framework comprising a large language model (LLM) with two distinct sets of LoRA parameters. The one LLM serves as a Decision Expert for scenario reasoning and driving decision-making, while the other acts as an Action Expert that dynamically maps linguistic decisions into feasible trajectories. By feeding trajectory-level rewards back into the reasoning space, MindDrive enables trial-and-error learning over a finite set of discrete linguistic driving decisions, instead of operating directly in a continuous action space. This approach effectively balances optimal decision-making in complex scenarios, human-like driving behavior, and efficient exploration in online reinforcement learning. Using the lightweight Qwen-0.5B LLM, MindDrive achieves Driving Score (DS) of 78.04 and Success Rate (SR) of 55.09% on the challenging Bench2Drive benchmark. To the best of our knowledge, this is the first work to demonstrate the effectiveness of online reinforcement learning for the VLA model in autonomous driving.
作者：Haoyu Fu, Diankun Zhang, Zongchuang Zhao, Jianfeng Cui, Hongwei Xie, Bing Wang, Guang Chen, Hangjun Ye, Dingkang Liang, Xiang Bai
来源
本页事实对应的公开页面，以原文为准。