CBDES MoE: Hierarchically Decoupled Mixture-of-Experts for Functional Modules in Autonomous Driving

TL;DR

This paper introduces CBDES MoE, a hierarchical mixture-of-experts architecture for autonomous driving perception systems, which improves adaptability, modeling capacity, and generalization by dynamically selecting experts at the functional module level.

Contribution

It presents the first modular Mixture-of-Experts framework at the functional module level for autonomous driving, integrating heterogeneous experts with a lightweight gating mechanism for efficient, input-aware inference.

Findings

CBDES MoE outperforms fixed single-expert baselines in 3D object detection.

Achieves 1.6-point higher mAP and 4.1-point higher NDS over the strongest single-expert model.

Demonstrates improved adaptability and generalization in real-world autonomous driving datasets.

Abstract

Bird's Eye View (BEV) perception systems based on multi-sensor feature fusion have become a fundamental cornerstone for end-to-end autonomous driving. However, existing multi-modal BEV methods commonly suffer from limited input adaptability, constrained modeling capacity, and suboptimal generalization. To address these challenges, we propose a hierarchically decoupled Mixture-of-Experts architecture at the functional module level, termed Computing Brain DEvelopment System Mixture-of-Experts (CBDES MoE). CBDES MoE integrates multiple structurally heterogeneous expert networks with a lightweight Self-Attention Router (SAR) gating mechanism, enabling dynamic expert path selection and sparse, input-aware efficient inference. To the best of our knowledge, this is the first modular Mixture-of-Experts framework constructed at the functional module granularity within the autonomous driving domain. Extensive evaluations on the real-world nuScenes dataset demonstrate that CBDES MoE consistently outperforms fixed single-expert baselines in 3D object detection. Compared to the strongest single-expert model, CBDES MoE achieves a 1.6-point increase in mAP and a 4.1-point improvement in NDS, demonstrating the effectiveness and practical advantages of the proposed approach.