On-Board Vision-Language Models for Personalized Autonomous Vehicle Motion Control: System Design and Real-World Validation

TL;DR

This work proposes a lightweight yet effective on-board VLM framework that provides low-latency personalized driving performance while maintaining strong reasoning capabilities and incorporates a Retrieval-Augmented Generation (RAG)-based memory module that enables continuous learning of individual driving preferences through human feedback.

Abstract

Personalized driving refers to an autonomous vehicle's ability to adapt its driving behavior or control strategies to match individual users' preferences and driving styles while maintaining safety and comfort standards. However, existing works either fail to capture every individual preference precisely or become computationally inefficient as the user base expands. Vision-Language Models (VLMs) offer promising solutions to this front through their natural language understanding and scene reasoning capabilities. In this work, we propose a lightweight yet effective on-board VLM framework that provides low-latency personalized driving performance while maintaining strong reasoning capabilities. Our solution incorporates a Retrieval-Augmented Generation (RAG)-based memory module that enables continuous learning of individual driving preferences through human feedback. Through comprehensive real-world vehicle deployment and experiments, our system has demonstrated the ability to provide safe, comfortable, and personalized driving experiences across various scenarios and significantly reduce takeover rates by up to 76.9%. To the best of our knowledge, this work represents the first end-to-end VLM-based motion control system in real-world autonomous vehicles.

Figures (6)

• Figure 1: An overview of the proposed framework for personalized autonomous vehicle motion control. The system workflow begins with processing four input streams (System Message $S$, Human Instruction $I$, Camera Image $V$, and Historical Memory $H$) through an on-board VLM, which generates personalized action policies $P$ containing MPC and PID control parameter matrices. These policies are then executed through the vehicle's drive-by-wire system. If the human evaluates, human feedback $F$ is collected and stored in the RAG-based memory module for continuous learning and adaptation of the system's behavior to individual preferences.
• Figure 2: Overview of the experiment setup.
• Figure 3: Sample vision inputs from different weather scenarios.
• Figure 4: Takeover rate comparison between the baseline and our method.
• Figure 5: Median comparison between VLM-based model and the baseline under various driving scenarios. Orange represents baseline while Blue represents our method.