The Method of Creel Positioning Based on Monocular Vision

Area of Science:

• Textile Engineering
• Computer Vision
• Robotics

Background:

• Automatic bobbin replacement in the textile industry faces challenges with bobbin position offset, loosening, and deformation, leading to replacement failures.
• Manual positioning is labor-intensive and unreliable, necessitating automated solutions for creel coordinate initialization.

Purpose of the Study:

• To propose and evaluate an automatic creel positioning method based on monocular vision to address inaccuracies in textile bobbin replacement.
• To improve the reliability and efficiency of the automated bobbin changing process by precisely determining creel coordinates.

Main Methods:

• An industrial camera was mounted on a truss-controlled manipulator to inspect yarn frames.
• An improved Hough circle detection algorithm was used to identify creel center coordinates (x, y) and radius (r) from real-time images.
• Camera calibration and specially designed creel positioning markers were employed to reduce environmental interference and enhance accuracy.

Main Results:

• The monocular vision method achieved fine positioning effects within a camera-to-creel distance of 170-190 mm.
• The best positioning performance was observed at 190 mm, yielding an average error of only 0.51 mm.
• The method demonstrated minimal deviation in center coordinates and radius, exceeding bobbin yarn changing accuracy requirements.

Conclusions:

• The proposed automatic creel positioning method based on monocular vision is effective for the textile industry.
• The system significantly improves positioning accuracy and reliability compared to manual methods.
• This technology offers a viable solution for enhancing automated bobbin replacement processes.