Scholarlink

<p>Human–robot collaborative systems face critical challenges in multi-task dynamic switching, precise contact force control, and safety constraint satisfaction during the digital transformation of manufacturing. However, existing methods typically address impedance control optimization and task planning as independent problems, lacking a unified framework that simultaneously ensures task switching smoothness, real-time adaptive control, and strict safety guarantees. To bridge this gap, this paper proposes a hierarchical quadratic programming-based impedance adaptive control method. The method constructs a three-layer control architecture of perception decision execution, deeply integrating the hierarchical quadratic programming framework with adaptive impedance control. It employs null space projection technology to ensure a strict guarantee of task priorities, designs a smooth switching mechanism based on activation functions to ensure continuity of task transitions, introduces hierarchical barrier functions to handle multi-level safety constraints, and implements online optimization of impedance parameters through a hybrid learning strategy combining gradient descent and recursive least squares. Experimental validation based on the UR10 collaborative robot platform demonstrates that: Under single-task steady-state conditions, the system achieves position control accuracy of 0.52&#xa0;mm and force tracking error of 1.23N; under comprehensive multi-task conditions, position RMSE reaches 0.65&#xa0;mm and force tracking error is 1.35N; multi-task switching response time is 0.35&#xa0;s with maximum overshoot of 4.2%; safety constraint satisfaction rate reaches 99.7%, with all contact force indicators more than 36% below ISO 15066 standard limits; compared with existing methods, comprehensive control accuracy is improved by 28.6% and computational efficiency is enhanced by 43.8%. Ablation experiments further confirm that each component—adaptive impedance learning, smooth switching, and barrier functions—makes a distinct and necessary contribution to overall performance. The research outcomes provide an efficient and safe control solution for human–robot collaboration in multi-task scenarios, with significant application value in intelligent manufacturing and flexible production fields.</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

Research on hierarchical quantitative impedance adaptive control method for human–robot collaborative systems in multi-task scenarios

  • 期刊论文

  • 发表日期：

    2026年5月23日

查看全文
  • Chao Li

摘要

Human–robot collaborative systems face critical challenges in multi-task dynamic switching, precise contact force control, and safety constraint satisfaction during the digital transformation of manufacturing. However, existing methods typically address impedance control optimization and task planning as independent problems, lacking a unified framework that simultaneously ensures task switching smoothness, real-time adaptive control, and strict safety guarantees. To bridge this gap, this paper proposes a hierarchical quadratic programming-based impedance adaptive control method. The method constructs a three-layer control architecture of perception decision execution, deeply integrating the hierarchical quadratic programming framework with adaptive impedance control. It employs null space projection technology to ensure a strict guarantee of task priorities, designs a smooth switching mechanism based on activation functions to ensure continuity of task transitions, introduces hierarchical barrier functions to handle multi-level safety constraints, and implements online optimization of impedance parameters through a hybrid learning strategy combining gradient descent and recursive least squares. Experimental validation based on the UR10 collaborative robot platform demonstrates that: Under single-task steady-state conditions, the system achieves position control accuracy of 0.52 mm and force tracking error of 1.23N; under comprehensive multi-task conditions, position RMSE reaches 0.65 mm and force tracking error is 1.35N; multi-task switching response time is 0.35 s with maximum overshoot of 4.2%; safety constraint satisfaction rate reaches 99.7%, with all contact force indicators more than 36% below ISO 15066 standard limits; compared with existing methods, comprehensive control accuracy is improved by 28.6% and computational efficiency is enhanced by 43.8%. Ablation experiments further confirm that each component—adaptive impedance learning, smooth switching, and barrier functions—makes a distinct and necessary contribution to overall performance. The research outcomes provide an efficient and safe control solution for human–robot collaboration in multi-task scenarios, with significant application value in intelligent manufacturing and flexible production fields.