多无人系统协同中的人工智能安全探索  

Artificial Intelligence Security in Multiple-Unmanned System Cooperation

作　　者：施文 王楷文 俞成浦 孙健[1] 陈杰[1,2] Shi Wen;Wang Kaiwen;Yu Chengpu;Sun Jian;Chen Jie(School of Automation,Beijing Institute of Technology,Beijing 100081,China;Tongji University,Shanghai 200092,China)

机构地区：[1]北京理工大学自动化学院,北京100081 [2]同济大学,上海200092

出　　处：《中国工程科学》2021年第3期82-89,共8页Strategic Study of CAE

基　　金：中国工程院咨询项目“新一代人工智能安全与自主可控发展战略研究”(2019-ZD-01)。

摘　　要：作为中国新一代人工智能规划中的重要组成部分,多无人系统协同是我国未来国防建设和社会发展的一项变革性技术。虽然多无人系统协同技术研究与系统集成已经达到了空前高度,但是其相关人工智能安全问题研究还处在萌芽阶段。本文阐述了统筹推进多无人系统协同赋能应用与风险防控的重大意义,提出了“四位一体”全面推进多无人系统协同安全发展的战略思路,探索了多无人系统协同在内生安全和衍生安全层面潜在的挑战与应对思路。研究提出了智能无人系统安全对策建议:构建国家级无人系统验证平台,推动人才队伍建设;逐步深化无人系统产业“放管服”,发展新一代人工智能安全生态;充分发挥多无人系统协同的优势,赋能保障和改善民生,服务构建人类命运共同体。Multiple unmanned system cooperation is an important component of China’s new-generation artificial intelligence(AI)planning and it is a transformative technology for future national defense construction and social development of China.Although significant achievements have been made regarding the technological research and system integration of multiple unmanned system cooperation,research on AI security is still in its infancy.Promoting multiple unmanned system cooperation enables AI application and risk control.In this article,we propose a four-in-one strategy to promote the collaborative and secure development of multiple unmanned systems,investigate potential challenges and countermeasures for multiple unmanned system cooperation at the endogenous and derivative security levels,and propose several suggestions regarding the security of intelligent unmanned systems.Specifically,a national unmanned system verification platform should be established to promote the construction of talent teams;services need to be upgraded for the unmanned system industry to develop a new-generation AI security ecology;and the advantages of multiple unmanned system cooperation should be maximized to improve people’s livelihood.

关 键 词：多无人系统协同 人工智能安全 安全风险防控 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]