端到端自动驾驶可直接将原始传感器输入映射为车辆底层控制指令，是具身智能（Embodied AI）的重要组成部分。尽管多模态大语言模型（Multimodal Large Language Models, MLLMs）在高层交通场景语义理解方面已取得成功，但如何将这些概念性语义理解有效转化为底层运动控制指令，并实现跨场景驾驶的泛化性与一致性，仍是一大挑战。我们提出了Sce2DriveX，这是一种类人驾驶思维链（chain-of-thought, CoT）推理的MLLM框架。Sce2DriveX利用局部场景视频与全局鸟瞰图（BEV）的多模态联合学习，深度理解长程时空关系与道路拓扑结构，增强其在3D动静态场景中的综合感知与推理能力，实现跨场景的驾驶泛化。在此基础上，该框架重构了人类驾驶中固有的内隐认知链，涵盖场景理解、元动作推理、行为解释分析、运动规划与控制，从而进一步缩小自动驾驶与人类思维过程之间的差距。为提升模型性能，我们构建了首个面向3D空间理解与长轴任务推理的大规模视觉问答（Visual Question Answering, VQA）驾驶指令数据集。大量实验表明，Sce2DriveX在从场景理解到端到端驾驶的任务上均达到了当前最优（state-of-the-art）性能，并且在CARLA Bench2Drive基准测试上展现出强大的泛化能力。
英文摘要：
End-to-end autonomous driving, which directly maps raw sensor inputs to low-level vehicle controls, is an important part of Embodied AI. Despite successes in applying Multimodal Large Language Models (MLLMs) for high-level traffic scene semantic understanding, it remains challenging to effectively translate these conceptual semantics understandings into low-level motion control commands and achieve generalization and consensus in cross-scene driving. We introduce Sce2DriveX, a human-like driving chain-of-thought (CoT) reasoning MLLM framework. Sce2DriveX utilizes multimodal joint learning from local scene videos and global BEV maps to deeply understand long-range spatiotemporal relationships and road topology, enhancing its comprehensive perception and reasoning capabilities in 3D dynamic/static scenes and achieving driving generalization across scenes. Building on this, it reconstructs the implicit cognitive chain inherent in human driving, covering scene understanding, meta-action reasoning, behavior interpretation analysis, motion planning and control, thereby further bridging the gap between autonomous driving and human thought processes. To elevate model performance, we have developed the first extensive Visual Question Answering (VQA) driving instruction dataset tailored for 3D spatial understanding and long-axis task reasoning. Extensive experiments demonstrate that Sce2DriveX achieves state-of-the-art performance from scene understanding to end-to-end driving, as well as robust generalization on the CARLA Bench2Drive benchmark.
D3VL: Understanding Driving Scenes from 3D Time Series Data and Video with Language Models
Bradley Department of Electrical and Computer Engineering, Virginia Tech（弗吉尼亚理工大学布拉德利电气与计算机工程系）；Virginia Tech Transportation Institute（弗吉尼亚理工大学交通研究所）；Sanghani Center for Artificial Intelligence and Data Analytics（桑哈尼人工智能与数据分析中心）
Scenes as an Attack Surface: Uncovering Scene-Grounded Safety Failures in MLLMs
Nanjing University（南京大学）；The Hong Kong University of Science and Technology (Guangzhou)（香港科技大学（广州））；China Academy of Launch Vehicle Technology（中国运载火箭技术研究院）；Beihang University（北京航空航天大学）
Evolution of Accuracy and Visual-Cognitive Errors in a Decade of Vision-Language AI Models
Psychological & Brain Sciences, University of California, Santa Barbara（加利福尼亚大学圣巴巴拉分校心理与脑科学系）；Department of Computer Science, University of California, Santa Barbara（加利福尼亚大学圣巴巴拉分校计算机科学系）；Department of Electrical and Computer Engineering, University of California, Santa Barbara（加利福尼亚大学圣巴巴拉分校电气与计算机工程系）
Seeing vs. Believing: Evaluating the Language Bias of Open-Source MLLMs in Counter-Intuitive Scenes
Zhejiang University（浙江大学）；Beijing University of Posts and Telecommunications（北京邮电大学）；Hong Kong University of Science and Technology (Guangzhou)（香港科技大学（广州））
MLLM-SUL: Multimodal Large Language Model for Semantic Scene Understanding and Localization in Traffic Scenarios
Shanghai Research Institute for Intelligent Autonomous Systems, Tongji University（同济大学智能自主系统研究院）；School of Automotive Studies, Tongji University（同济大学汽车学院）；School of Mechanical Engineering, Shanghai Jiao Tong University（上海交通大学机械与动力工程学院）
Towards Unified Surgical Scene Understanding:Bridging Reasoning and Grounding via MLLMs
Southern University of Science and Technology（南方科技大学）；Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences（深圳先进技术研究院，中国科学院）；Northwestern University（西北大学）；University of Alberta（阿尔伯塔大学）；Yale University（耶鲁大学）；Nanfang Hospital（南华医院）；Shenzhen University of Advanced Technology（深圳大学先进技术研究院）
ViSTR-Bench: Can MLLMs Reason from Continuous Visual Cues in Dynamic Scenes?
School of Artificial Intelligence, Beihang University（北京航空航天大学人工智能学院）；Zhongguancun Academy（中关村科学城）；School of Engineering, Westlake University（西湖大学工学院）；School of Transportation Science and Engineering, Beihang University（北京航空航天大学交通科学与工程学院）
Efficient Physics Simulation for 3D Scenes via MLLM-Guided Gaussian Splatting
School of Computer Science, Wuhan University（武汉大学计算机学院）；Wuhan National Laboratory for Optoelectronics, Huazhong University of Science and Technology（华中科技大学光电研究院）；Meta Reality Lab（Meta现实实验室）；Xi’an Jiao Tong University（西安交通大学）；National University of Singapore（新加坡国立大学）；The Department of Systems Hub, Hong Kong University of Science and Technology (Guangzhou)（香港科技大学系统枢纽部门（广州））
TriViewBench: Controlled Complexity Scaling for Multi-View Structural Reasoning in MLLMs
School of Intelligence Science and Technology, Nanjing University（南京大学智能科学与技术学院）；National Key Laboratory for Novel Software Technology, Nanjing University（南京大学计算机软件新技术国家重点实验室）
Reinforcing Egocentric Spatial Perception in Multimodal Large Language Models via Ego Scene Augmentation
The Hong Kong University of Science and Technology (Guangzhou)（香港科技大学（广州））；Guangxi Zhuang Autonomous Region Information Center（广西壮族自治区信息中心）；The Hong Kong University of Science and Technology（香港科技大学）
I Spy With My Model's Eye: Visual Search as a Behavioural Test for MLLMs
Leverhulme Centre for the Future of Intelligence, University of Cambridge（未来智能研究中心、剑桥大学）；Department of Engineering, University of Cambridge（工程系、剑桥大学）；Department of Psychology, University of Cambridge（心理学系、剑桥大学）；Department of Computer Science, University of Cambridge（计算机科学系、剑桥大学）
HumanPCR: Probing MLLM Capabilities in Diverse Human-Centric Scenes
Key Laboratory of Intelligent Information Processing of Chinese Academy of Sciences (CAS), Institute of Computing Technology, CAS, China（中国科学院智能信息处理重点实验室（中国科学院）、计算技术研究所、中国）
Turning Generators into Retrievers: Unlocking MLLMs for Natural Language-Guided Geo-Localization
Vermont Artificial Intelligence Lab, Department of Computer Science, University of Vermont（佛蒙特大学计算机科学系佛蒙特人工智能实验室）；Intelligent Machines Lab, Department of Artificial Intelligence, Information Technology University（信息技术大学人工智能系智能机器实验室）；Institute of Artificial Intelligence, University of Central Florida（中佛罗里达大学人工智能研究所）
Follow-Your-Instruction: A Comprehensive MLLM Agent for World Data Synthesis
HKUST(GZ)（香港科技大学（广州））；HKUST（香港科技大学）；Tsinghua Univerisity（清华大学）；Peking University（北京大学）；Chongqing University（重庆大学）；Beijing Innovation Center of Humanoid Robotics（北京人形机器人创新中心）
EmbodiedVSR: Dynamic Scene Graph-Guided Chain-of-Thought Reasoning for Visual Spatial Tasks
Beijing Innovation Center of Humanoid Robotics（北京人形机器人创新中心）；Hong Kong University of Science and Technology (Guangzhou)（香港科技大学（广州））；Hong Kong University of Science and Technology（香港科技大学）；University of Technology Sydney（悉尼科技大学）
