Scholarlink

<p>The Automatic Milking System has significantly advanced dairy farming by automating milking procedures, yet current solutions often address operational challenges like positioning and force perception in isolation, lacking an integrated approach to ensure both operational safety and structural reliability. To address these issues, this paper introduces a Hierarchical Closed-loop Digital Twin Modeling paradigm that establishes a multi-layer digital twin system for milking robots. The core of the paradigm is its bidirectional, closed-loop modeling at both the static and interaction layers. At the static layer, a deep neural network-based finite element surrogate model is developed for load inversion and stress prediction. At the interaction layer, a Cosserat-based hybrid model is employed for real-time contact force estimation in flexible fingers. Experimental results demonstrate that the proposed system achieves high accuracy in load prediction (relative error &lt; 5.62%) and contact force control (relative error &lt; 2.76% vs. the 15N threshold). The results confirm that the Hierarchical Closed-loop Digital Twin Modeling paradigm effectively integrates structural and interactive safety functions, providing a scalable and systematic modeling reference for developing complex agricultural robotic systems.</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

A digital twin for enhanced safety and control in milking robots: a hierarchical closed-loop modeling paradigm

  • 期刊论文

  • 发表日期：

    2026年3月18日

查看全文
  • Pengyu Wang,
  • Guohua Gao,
  • Yongbing Feng,
  • Yincheng Lv

摘要

The Automatic Milking System has significantly advanced dairy farming by automating milking procedures, yet current solutions often address operational challenges like positioning and force perception in isolation, lacking an integrated approach to ensure both operational safety and structural reliability. To address these issues, this paper introduces a Hierarchical Closed-loop Digital Twin Modeling paradigm that establishes a multi-layer digital twin system for milking robots. The core of the paradigm is its bidirectional, closed-loop modeling at both the static and interaction layers. At the static layer, a deep neural network-based finite element surrogate model is developed for load inversion and stress prediction. At the interaction layer, a Cosserat-based hybrid model is employed for real-time contact force estimation in flexible fingers. Experimental results demonstrate that the proposed system achieves high accuracy in load prediction (relative error < 5.62%) and contact force control (relative error < 2.76% vs. the 15N threshold). The results confirm that the Hierarchical Closed-loop Digital Twin Modeling paradigm effectively integrates structural and interactive safety functions, providing a scalable and systematic modeling reference for developing complex agricultural robotic systems.