DriveMoE: Mixture-of-Experts for Vision-Language-Action Model in End-to-End Autonomous Driving

TL;DR

DriveMoE introduces a Mixture-of-Experts framework for end-to-end autonomous driving, enhancing scenario handling by dynamically selecting relevant visual cues and behaviors, achieving state-of-the-art results.

Contribution

This work pioneers the integration of Scene-Specialized Vision MoE and Skill-Specialized Action MoE into an end-to-end autonomous driving model, inspired by human cognition.

Findings

DriveMoE achieves state-of-the-art performance on Bench2Drive evaluation.

Dynamic routing improves handling of diverse and rare driving scenarios.

Explicit behavioral specialization prevents modes averaging issues.

Abstract

End-to-end autonomous driving (E2E-AD) demands effective processing of multi-view sensory data and robust handling of diverse and complex driving scenarios, particularly rare maneuvers such as aggressive turns. Recent success of Mixture-of-Experts (MoE) architecture in Large Language Models (LLMs) demonstrates that specialization of parameters 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. DriveMoE is built upon our $\pi_0$ Vision-Language-Action (VLA) baseline (originally from the embodied AI field), called Drive-$\pi_0$. Specifically, we add Vision MoE to Drive-$\pi_0$ by training a router to select relevant cameras according to the driving context dynamically. This design mirrors human driving cognition, where drivers selectively attend to crucial visual cues rather than exhaustively processing all visual information. In addition, we add Action MoE by training another router to activate specialized expert modules for different driving behaviors. Through explicit behavioral specialization, DriveMoE is able to handle diverse scenarios without suffering from modes averaging like existing models. In Bench2Drive closed-loop evaluation experiments, DriveMoE achieves state-of-the-art (SOTA) performance, demonstrating the effectiveness of combining vision and action MoE in autonomous driving tasks. We will release our code and models of DriveMoE and Drive-$\pi_0$.