复杂环境下自动驾驶系统软件虚拟测试与验证方法  

Virtual Testing and Validation Approach for Autonomous Driving Software In Complex Environments

作　　者：胡骏[1] 胡天翼 Jun Hu;Tianyi Hu(Beijing Institute of Computer Application and Simulation Technology,Beijing 100854;CIeNET Technologies,Beijing 100102)

机构地区：[1]北京计算机应用和仿真技术研究所,北京100854 [2]瞬联软件科技(北京)有限公司,北京100102

出　　处：《汽车文摘》2019年第10期1-7,共7页Automotive Digest

摘　　要：自动驾驶系统是一个复杂的智能化系统,对于这类系统和软件的测试与验证面临巨大的挑战,目前更多依赖在实际路况下的测试,除路试道路和环境条件受限外,需要耗费大量人力时间成本,而且对于不同路况、天气、相邻静止和运动实体的复杂环境,特别对一些突发状况难以人工设置和复制,导致系统测试与验证不可重复。探讨针对复杂环境开展自动驾驶系统软件虚拟测试与验证的方法,给出了车载和实验室仿真测试与验证系统的组成结构和相关实现方法,以克服在实际路况下测试的局限性,在路试基础上进一步提高软件测试充分性,从而达到提高系统安全性和可靠性的目的。The automated driving system for autonomous vehicle is a complex intelligent system, which is facing the huge challenges for testing and validating the system and software. Currently it mainly relies on testing under actual road conditions which costs lot of manpower and time in addition to the limitations of road and environmental conditions for road testing. Furthermore the complex environments for different road conditions, weather, adjacent stationary and moving entities, especially for some unexpected situations, are difficult to set and restored manually, that causes system test and validation to be non-repeatable. This paper discusses virtual testing and validation approach for the automated driving system software in complex environments and presents the structures of on- board and laboratory simulation test and validation systems and related implementation methods in order to overcome the limitations, further improve the sufficiency of software testing based on road testing and thus achieve the purpose of improving system security and reliability.

关 键 词：自主系统 自动驾驶 软件测试 虚拟测试 仿真环境 

分 类 号：U463.6[机械工程—车辆工程]