MiniDrive: More Efficient Vision-Language Models with Multi-Level 2D Features as Text Tokens for Autonomous Driving

TL;DR

MiniDrive introduces a lightweight vision-language framework for autonomous driving that efficiently processes multi-level 2D features, outperforming previous models in speed and parameter size while maintaining high accuracy.

Contribution

The paper presents MiniDrive, a novel framework with FE-MoE and DI-Adapter modules that enhance efficiency and multi-image processing in vision-language models for autonomous driving.

Findings

Achieves state-of-the-art performance with only 83M parameters.

Reduces computational cost and improves response efficiency.

Effectively handles multi-camera perception tasks.

Abstract

Vision-language models (VLMs) serve as general-purpose end-to-end models in autonomous driving, performing subtasks such as prediction, planning, and perception through question-and-answer interactions. However, most existing methods rely on computationally expensive visual encoders and large language models (LLMs), making them difficult to deploy in real-world scenarios and real-time applications. Meanwhile, most existing VLMs lack the ability to process multiple images, making it difficult to adapt to multi-camera perception in autonomous driving. To address these issues, we propose a novel framework called MiniDrive, which incorporates our proposed Feature Engineering Mixture of Experts (FE-MoE) module and Dynamic Instruction Adapter (DI-Adapter). The FE-MoE effectively maps 2D features into visual token embeddings before being input into the language model. The DI-Adapter enables the visual token embeddings to dynamically change with the instruction text embeddings, resolving the issue of static visual token embeddings for the same image in previous approaches. Compared to previous works, MiniDrive achieves state-of-the-art performance in terms of parameter size, floating point operations, and response efficiency, with the smallest version containing only 83M parameters.