Robotic Threaded Insert Assembly With LSTM Motion State Recognition and Stable Insertion Control
Sijie Peng, Zhuchen Han, Junpeng Chen, Zijing Zeng, Donghao Li, Yang Pan
Abstract
Threaded insert assembly is a critical process in automotive manufacturing, with its installation quality directly determining the structural integrity of vehicle components. Due to the complexity of the industrial scenarios, threaded insert assembly faces two major challenges: peg-hole misalignment due to visual errors and dynamic contact forces from interference fit during insertion. To address these issues, this study proposes a novel threaded insert assembly strategy that significantly enhances both operational efficiency and stability. First, a force controlled search method guided by motion state recognition is developed to achieve fast searching. Next, a stable insertion control method with online adaptive notch filtering is utilized for safe insert installation. Experimental results demonstrate the robust generalizability of the recognition model across threaded workpieces, reducing the average positional error by over 85% after the searching phase. Furthermore, the proposed method reduces the maximum contact force by over 50% during the inserting phase. Experiments on automotive chassis workpieces demonstrate that the proposed strategy enables stable and efficient automatic assembly of threaded inserts under industrial conditions.
BibTeX
@inproceedings{ral2026_roboticthreadedi,
title = {Robotic Threaded Insert Assembly With LSTM Motion State Recognition and Stable Insertion Control},
author = {Sijie Peng and Zhuchen Han and Junpeng Chen and Zijing Zeng and Donghao Li and Yang Pan},
booktitle = {RA-L 2026},
year = {2026}
}