Scholarlink

<p>The commercialization of Level 3 (L3) conditionally automated vehicles introduces critical challenges in human-machine shared control. This paper proposes a Human-Machine Cyber-Physical System (HM-CPS) framework to investigate the evolutionary strategic choices between human drivers and autonomous driving algorithms using an evolutionary game model. We systematically analyze interactive behaviors under the influence of psychological risk compensation effects, legal liability allocation, and algorithmic computational costs. Theoretical analysis and numerical simulations reveal that extreme legal liability boundaries trigger “free-riding” behaviors, driving the system toward suboptimal equilibrium states where one party becomes overly reliant on the other. Furthermore, we demonstrate that high computational burdens force autonomous systems into aggressive states, compromising overall safety margins. These findings provide data-driven, quantitative guidelines for policymakers and automotive engineers to optimally calibrate liability frameworks and allocate onboard computing resources, ultimately fostering a safe and trustworthy human-machine co-driving ecosystem.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Evolutionary game analysis of liability boundaries in L3 autonomous driving: an electrical engineering perspective

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年6月16日

查看全文
  • Shikai Wang,
  • A. Rong

摘要

The commercialization of Level 3 (L3) conditionally automated vehicles introduces critical challenges in human-machine shared control. This paper proposes a Human-Machine Cyber-Physical System (HM-CPS) framework to investigate the evolutionary strategic choices between human drivers and autonomous driving algorithms using an evolutionary game model. We systematically analyze interactive behaviors under the influence of psychological risk compensation effects, legal liability allocation, and algorithmic computational costs. Theoretical analysis and numerical simulations reveal that extreme legal liability boundaries trigger “free-riding” behaviors, driving the system toward suboptimal equilibrium states where one party becomes overly reliant on the other. Furthermore, we demonstrate that high computational burdens force autonomous systems into aggressive states, compromising overall safety margins. These findings provide data-driven, quantitative guidelines for policymakers and automotive engineers to optimally calibrate liability frameworks and allocate onboard computing resources, ultimately fostering a safe and trustworthy human-machine co-driving ecosystem.