Scholarlink

<p>To enable efficient motion and stiffness generalization on resource-constrained platforms, this paper proposes a lightweight method called arc-normalized affine deformation (ANAD). The method constructs a unified representation of human-demonstrated trajectories and spatial stiffness profiles, and applies an arc-length-normalized affine deformation to achieve simultaneous adaptation. The influence of directional deformation strategies on motion consistency and compliance adaptation is analyzed. Unlike conventional methods that rely on complex parameter tuning or numerical integration, ANAD requires no parameter tuning, offering high interpretability and real-time performance with minimal computation. Experiments on KUKA and UR5e collaborative robots demonstrate that ANAD can accurately generalize both motion and stiffness. Compared with the traditional DMP method, ANAD reduces computation time by over 99 %, validating its effectiveness and practicality in real-time human–robot collaboration.</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

Simultaneous generalization of trajectory and stiffness based on arc-normalized affine deformation in human–robot collaboration

  • 期刊论文

  • 发表日期：

    2026年4月9日

查看全文
  • Yuankai Lu,
  • Yujie Zhang,
  • Haibin Wu

摘要

To enable efficient motion and stiffness generalization on resource-constrained platforms, this paper proposes a lightweight method called arc-normalized affine deformation (ANAD). The method constructs a unified representation of human-demonstrated trajectories and spatial stiffness profiles, and applies an arc-length-normalized affine deformation to achieve simultaneous adaptation. The influence of directional deformation strategies on motion consistency and compliance adaptation is analyzed. Unlike conventional methods that rely on complex parameter tuning or numerical integration, ANAD requires no parameter tuning, offering high interpretability and real-time performance with minimal computation. Experiments on KUKA and UR5e collaborative robots demonstrate that ANAD can accurately generalize both motion and stiffness. Compared with the traditional DMP method, ANAD reduces computation time by over 99 %, validating its effectiveness and practicality in real-time human–robot collaboration.