Scholarlink

<p>Robotic assembly of avionics connectors is one important and challenging peg-in-hole task for industrial automation. Searching for holes is one extremely crucial step for this type of assembly, and its operation is generally based on contact state recognition. Conventional methods for identifying contact states usually employ the force/torque feedback at particular moments, which may reduce the feature information of contact state recognition. Although improved identifying strategies of contact states based on sequential force/torque or potential field may enhance the feature recognition capability in a certain degree, they require the large adjustment of orientation at the end, which may cause damages to the electrical components especially for avionics connectors with complex structures. So far, effective robotic assembly method has not been reported for the avionics connectors. Therefore, the objective of this study is to propose one automatic assembly strategy by combining sequential force/torque and long short-term memory (LSTM) to gain more feature information and protect the electrical components well. Furthermore, the searching path is planned based on one quarter of the Archimedean spiral; thus, the searching area is narrowed significantly, and the searching efficiency is improved greatly. Theoretical and experimental studies are conducted to validate the proposed method. The operation environment of the robotic test rig and the assembly strategy are presented at the beginning. The initial positioning is carried out based on the vision identification. Then, the contact state recognition is conducted and the improved searching path is proposed. Subsequently, the impedance control method is employed for the inserting procedure. Taking five different shapes and sizes of avionics connectors as assembly objects, the experimental testing results show that the accuracy of each contact state recognition model is greater than 95.8%, and the assembly success rate is greater than 94%. Compared with the conventional assembly strategy without contact state recognition, the assembly efficiency is increased by 66.7%.</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

An improved searching strategy based on contact state recognition for the assembly of avionics connector

  • 期刊论文

  • 发表日期：

    2025年6月20日

查看全文
  • Nannan Du,
  • Liang Yan,
  • Pengjie Xiang,
  • Xiaoshan Gao,
  • Xinghua He,
  • Suwan Bu,
  • I.-Ming Chen

摘要

Robotic assembly of avionics connectors is one important and challenging peg-in-hole task for industrial automation. Searching for holes is one extremely crucial step for this type of assembly, and its operation is generally based on contact state recognition. Conventional methods for identifying contact states usually employ the force/torque feedback at particular moments, which may reduce the feature information of contact state recognition. Although improved identifying strategies of contact states based on sequential force/torque or potential field may enhance the feature recognition capability in a certain degree, they require the large adjustment of orientation at the end, which may cause damages to the electrical components especially for avionics connectors with complex structures. So far, effective robotic assembly method has not been reported for the avionics connectors. Therefore, the objective of this study is to propose one automatic assembly strategy by combining sequential force/torque and long short-term memory (LSTM) to gain more feature information and protect the electrical components well. Furthermore, the searching path is planned based on one quarter of the Archimedean spiral; thus, the searching area is narrowed significantly, and the searching efficiency is improved greatly. Theoretical and experimental studies are conducted to validate the proposed method. The operation environment of the robotic test rig and the assembly strategy are presented at the beginning. The initial positioning is carried out based on the vision identification. Then, the contact state recognition is conducted and the improved searching path is proposed. Subsequently, the impedance control method is employed for the inserting procedure. Taking five different shapes and sizes of avionics connectors as assembly objects, the experimental testing results show that the accuracy of each contact state recognition model is greater than 95.8%, and the assembly success rate is greater than 94%. Compared with the conventional assembly strategy without contact state recognition, the assembly efficiency is increased by 66.7%.