Scholarlink

<p>Human-machine interaction (HMI) is a key component of conditional autonomous driving. However, the HMI system based on the steering wheel module often suffers from delayed recognition of takeover signals, potential actuator faults, and jerkiness during the takeover process. To address these challenges, this paper proposes a seamless HMI architecture combined with intention recognition and fault-tolerant control. First, a composite sensor capable of simultaneously acquiring triboelectric and pressure signals is designed, adopting a shared structural layer to minimize thickness, and an associated takeover intention recognition scheme is developed. Then, a four-dimensional hand wheel model that captures the transient characteristics of the steering wheel is proposed, based on which a switchable tube-model predictive control (MPC) controller combined with an auxiliary sliding-mode control law is constructed. Finally, an adaptive weight adjustment function is integrated into the tube-MPC framework to ensure a smooth takeover process. Experimental results demonstrate that the proposed architecture enables seamless HMI, including accurate driver intention recognition and promising tracking accuracy under both healthy and faulty conditions.</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

Seamless human-machine interaction via intention recognition based on triboelectric-pressure signals and fault-tolerant control with switchable torque-angle targets

  • 期刊论文

  • 发表日期：

    2026年8月3日

查看全文
  • Xiang Chen,
  • Yongyao Yuan,
  • Yicai Liu,
  • Xiangyu Wang,
  • Yushu Li,
  • Liang Li,
  • Wanzhong Zhao

摘要

Human-machine interaction (HMI) is a key component of conditional autonomous driving. However, the HMI system based on the steering wheel module often suffers from delayed recognition of takeover signals, potential actuator faults, and jerkiness during the takeover process. To address these challenges, this paper proposes a seamless HMI architecture combined with intention recognition and fault-tolerant control. First, a composite sensor capable of simultaneously acquiring triboelectric and pressure signals is designed, adopting a shared structural layer to minimize thickness, and an associated takeover intention recognition scheme is developed. Then, a four-dimensional hand wheel model that captures the transient characteristics of the steering wheel is proposed, based on which a switchable tube-model predictive control (MPC) controller combined with an auxiliary sliding-mode control law is constructed. Finally, an adaptive weight adjustment function is integrated into the tube-MPC framework to ensure a smooth takeover process. Experimental results demonstrate that the proposed architecture enables seamless HMI, including accurate driver intention recognition and promising tracking accuracy under both healthy and faulty conditions.