DriveMoE：面向端到端自动驾驶视觉-语言-动作模型的混合专家方法
DriveMoE: Mixture-of-Experts for Vision-Language-Action Model in End-to-End Autonomous Driving
DriveMoE 将 MoE 引入端到端自动驾驶的 VLA 框架，其视觉 MoE 通过路由器自适应选择相关相机视角，动作 MoE 通过路由器激活驾驶行为专家，这种按需激活专家的思路可复用于多视角感知与多模态动作生成任务，缓解模式平均问题。
原文依据 we propose DriveMoE, a novel MoE-based E2E-AD framework, with a Scene-Specialized Vision MoE and a Skill-Specialized Action MoE.
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论端到端自动驾驶（E2E-AD）要求有效处理多视角传感器数据，并稳健应对多样且复杂的驾驶场景，尤其是激进转向等罕见操作。混合专家（MoE）架构在大语言模型（LLM）中的近期成功表明，专家专业化能够实现强大的可扩展性。在本工作中，我们提出 DriveMoE，一种新颖的基于 MoE 的端到端自动驾驶框架，包含场景专业化的视觉 MoE 和技能专业化的动作 MoE。首先，我们引入 Drive-p0，一个从具身智能适配到自动驾驶的视觉-语言-动作（VLA）基线，作为 DriveMoE 的基础模型。在此基础上，我们通过精心设计的视觉 MoE 增强感知，其中路由器自适应地选择与上下文相关的相机视角。该机制受人类驾驶认知启发，即注意力指向关键视觉线索而非同时处理所有感官输入。在感知之外，我们引入动作 MoE，通过训练路由器激活针对不同驾驶行为定制的专家模块来增强框架。在动作 MoE 中，我们实现了两种不同风格（Token 级路由器和轨迹级路由器），并广泛探索了它们在自动驾驶中的适用性。在 Bench2Drive 闭环评估中，DriveMoE 在多样驾驶场景中展现出稳健性能，缓解了限制现有模型的模式平均效应，并取得了最优结果，较 Drive-p0 有显著提升。我们将发布 DriveMoE 和 Drive-p0 的代码与模型。
英文原文摘要 1661 字符
End-to-end autonomous driving (E2E-AD) demands effective processing of multi-view sensor data and robust handling of diverse and complex driving scenarios, particularly rare maneuvers such as aggressive turns. The recent success of the Mixture-of-Experts (MoE) architecture in Large Language Models (LLMs) demonstrates that expert specialization enables strong scalability. In this work, we propose DriveMoE, a novel MoE-based E2E-AD framework, with a Scene-Specialized Vision MoE and a Skill-Specialized Action MoE. First, we introduce Drive-p0, a Vision-Language-Action (VLA) baseline adapted from Embodied AI for autonomous driving, which serves as the foundation model for DriveMoE. Building on this, we strengthen perception through a carefully designed Vision MoE, where a router adaptively selects context-relevant camera views. This mechanism is inspired by human driving cognition, in which attention is directed to key visual cues rather than to all sensory inputs simultaneously. Beyond perception, we introduce an Action MoE that augments the framework by training a router to activate specialized expert modules tailored to distinct driving behaviors. Within the Action MoE, we implement two distinct styles(Token-level Router and Trajectory-level Router) and extensively explore their applicability in autonomous driving. In Bench2Drive closed-loop evaluations, DriveMoE demonstrates robust performance across diverse driving scenarios, alleviates the mode-averaging effect that limits existing models, and achieves state-of-the-art results with significant improvements over Drive-p0. We will release our code and models of DriveMoE and Drive-p0.