With the broader adoption and highly successful development of large language models (LLMs), there has been growing interest and demand for applying LLMs to autonomous driving technology. Driven by their natural language (NL) understanding and reasoning capabilities, LLMs have the potential to enhance various aspects of autonomous driving systems, from perception and scene understanding to interactive decision-making. This article first introduces the novel concept of designing LLMs for autonomous driving (LLM4AD), followed by a review of existing LLM4AD studies. Then, a comprehensive benchmark is proposed for evaluating the instruction-following and reasoning abilities of LLM4AD systems, which includes LaMPilot-Bench, CARLA Leaderboard 1.0 Benchmark in simulation and NuPlanQA for multiview visual question answering (VQA). Furthermore, extensive real-world experiments are conducted on autonomous vehicle platforms, examining both on-cloud and on-edge LLM deployment for personalized decision-making and motion control. Next, the future trends of integrating language diffusion models into autonomous driving are explored, exemplified by the proposed vision-language diffusion (ViLaD) framework. Finally, the main challenges of LLM4AD are discussed, including latency, deployment, security and privacy, safety, trust and transparency, and personalization.

Vision-Language Models (VLMs) have demonstrated significant promise for autonomous driving due to their powerful multimodal reasoning capabilities. However, adapting VLMs from generic data to safety-critical driving contexts introduces a notable challenge known as domain shift. Existing simulation-based and dataset-driven evaluation approaches struggle to accurately replicate real-world complexities, lacking repeatable closed-loop evaluation and flexible scenario manipulation. Furthermore, current real-world testing platforms typically focus on isolated modules and do not support comprehensive interaction with VLM-based systems. Consequently, there is a critical need for a holistic testing architecture capable of integrating perception, planning, and control modules, accommodating VLM-based systems, and supporting configurable real-world testing scenarios. In this article, we address this critical gap by proposing a hierarchical real-world test platform specialized in the rigorous evaluation of VLM-integrated autonomous driving systems. Specifically, our platform features have: a lightweight, structured, and low-latency middleware pipeline specialized for seamless VLM integration; a hierarchical modular architecture enabling flexible substitution between conventional and VLM-based autonomy components, providing exceptional deployment flexibility for rapid experimentation; and sophisticated closed-loop scenario-based testing capabilities on a controlled test track, facilitating comprehensive evaluation of the entire full-stack VLM-integrated autonomous driving pipeline, from perception, reasoning, decision-making, and planning to final vehicle maneuvers. Through an extensive real-world case study, we demonstrate the effectiveness of our platform in evaluating the performance and robustness of VLM-integrated autonomous driving under diverse realistic conditions. Project page and codes: https://github.com/YupengZhouPurdue/VLMTest .

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’s 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 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 personalized VLM motion control system in real-world autonomous vehicles. The demo video can be watched at https://tinyurl.com/4xsnz79n.

Collaborative perception empowers each agent to improve its perceptual ability through the exchange of perceptual messages with other agents. It inherently results in a fundamental tradeoff between perception ability and communication cost. To address this bottleneck issue, our core idea is to optimize the collaborative messages from two key aspects: representation and selection. The proposed codebook-based message representation enables the transmission of integer codes, rather than high-dimensional feature maps. The proposed information-filling-driven message selection optimizes local messages to collectively fill each agent's information demand, preventing information overflow among multiple agents. By integrating these two designs, we propose CodeFilling, a novel communication-efficient collaborative perception system, which significantly advances the perception-communication tradeoff and is inclusive to both homogeneous and heterogeneous collaboration settings. We evaluate CodeFilling in both a real-world dataset, DAIR-V2X, and a new simulation dataset, OPV2VH+. Results show that CodeFilling outperforms previous SOTA Where2comm on DAIR-V2X/OPV2VH+ with 1,333/1,206× lower communication volume. Our code is available at https://github.com/PhyllisH/CodeFilling.

To analyze multivariate time series, most previous methods assume regular subsampling of time series, where the interval between adjacent measurements and the number of samples remain unchanged. Practically, data collection systems could produce irregularly sampled time series due to sensor failures and interventions. However, existing methods designed for regularly sampled multivariate time series cannot directly handle irregularity owing to misalignment along both temporal and variate dimensions. To fill this gap, we propose Compatible Transformer (CoFormer), a transformer-based encoder to achieve comprehensive temporal-interaction feature learning for each individual sample in irregular multivariate time series. In CoFormer, we view each sample as a unique variate-time point and leverage intra-variate/inter-variate attentions to learn sample-wise temporal/interaction features based on intra-variate/inter-variate neighbors. With CoFormer as the core, we can analyze irregularly sampled multivariate time series for many downstream tasks, including classification and prediction. We conduct extensive experiments on 3 real-world datasets and validate that the proposed CoFormer significantly and consistently outperforms existing methods. Code will be avilable at https://github.com/MediaBrain-SJTU/CoFormer.