多模态信息融合在智能网联汽车交通事故鉴定中的应用探索  

Exploration of Multimodal Information Fusion in Traffic Accident Appraisal of Intelligent and Connected Vehicles

作　　者：冯浩 吴何坚 陈丰[1] 刘嘉鲯 FENG Hao;WU Hejian;CHEN Feng;LIU Jiaqi(Key Laboratory of Road and Traffic Engineering of the Ministry of Education,Tongji University,Shanghai 201804,China;Shanghai Forensic Service Platform,Key Laboratory of Forensic Science,Ministry of Justice,Academy of Forensic Science,Shanghai 200063,China;Key Laboratory of Intelligent Connected Vehicles Traffic Safety,Shanghai Public Security Bureau,Traffic Police Corps of Shanghai Public Security Bureau,Shanghai 201315,China)

机构地区：[1]同济大学道路与交通工程教育部重点实验室,上海201804 [2]司法鉴定科学研究院上海市司法鉴定专业技术服务平台司法部司法鉴定重点实验室,上海200063 [3]上海市公安局交通警察总队上海市公安局智能网联汽车交通安全研究重点实验室,上海201315

出　　处：《中国司法鉴定》2024年第5期18-26,共9页Chinese Journal of Forensic Sciences

基　　金：“十四五”国家重点研发计划(2023YFC3009703);上海市司法鉴定专业技术服务平台资助项目。

摘　　要：随着智能网联汽车产业快速发展,智能网联汽车的事故调查受到各地政府和主管部门的重视。首先,从交通事故鉴定的角度出发,针对智能网联汽车普遍存在的多源异构信息,从典型信息的执行标准、功能认证、车型范围、存储位置和获取途径等方面进行分析。其次,结合道路交通事故的调查取证规则和鉴定意见的证据要求,对智能网联汽车事故的典型信息进行溯源分析和证据力分析。再次,根据智能网联汽车事故调查目的和鉴定需求,结合其多源异构特点,提出多模态信息校验和融合。最后,对智能网联汽车的事故调查方法和事故鉴定技术进行展望。With the rapid development of the intelligent and connected vehicles industry,the accident investigation of intelligent and connected vehicles has attracted general attentions of local governments and authorities in China. Firstly, from the perspective of traffic accident analysis, the paper first analyzed the multi-source heterogeneous information prevalent in intelligent and connected vehicles, in terms of the excecution standards, functions authentication, scope of models, storage location and access to typical information. Secondly, combined with the rules of investigation and evidence collection in road traffic accidents and the evidence requirements of appraisal opinions, the typical information involved in intelligent and connected vehicles accidents was analyzed for traceability and evidence effectiveness. Thirdly, according to the purpose of intelligent and connected vehicles accidents investigation and appraisal requirements, combined with the multi-source heterogeneous characteristics of acquired information, the calibration and fusion of multi-modal information was proposed. Finally, an outlook on accident investigation and identification techniques for intelligent and connected vehicles was presented.

关 键 词：智能网联汽车 交通事故鉴定 多模态信息融合 

分 类 号：U491.3[交通运输工程—交通运输规划与管理]