Vision-based autonomous robot body alignment for copper wire spool pick up

• Main Authors: Daud, Mohd Razali, Shalfarina, Shahriman, Mohd Faudzi, Ahmad Afif, Sulaiman, Mohd Herwan, Irawan, Addie, Musa, Zulkifli
• Format: Research Report
• Language: English
• Published: 2019
• Subjects: TK Electrical engineering. Electronics Nuclear engineering
• Online Access: http://umpir.ump.edu.my/id/eprint/36354/1/Vision-based%20autonomous%20robot%20body%20alignment%20for%20copper%20wire%20spool%20pick%20up.wm.pdf; http://umpir.ump.edu.my/id/eprint/36354/

A simple, inexpensive system and effective in performing required tasks is the most preferable in industry. The peg-in-hole task is widely used in manufacturing process by using vision system and sensors but costly and needs complex algorithm. Picking up process of copper wire spools which are arranged side by side on a rack is also applying peg-in-hole concept. Currently, a forklift-like robot controlled using wired controllers is used. However, it is difficult for the operator to ensure the stem is properly inserted into the hole (peg-in-hole problem) because of the structure of the robot. However, the holder design is not universal and not applicable to other companies. The spool can only be grasped and pulled out from the front side and cannot be grasped using robot arm and gripper. In this study, a vision system is developed to solve the peg-in-hole problem by enabling the robot to autonomously perform the insertion and pick up the spool without using any sensors except a low-cost camera. A low-cost camera is used to capture images of copper wire spool in real-time video. Inspired by how human perceive an object orientation based on its shape, a system is developed to determine camera orientation based on the spool image condition and yaw angle from the center of the camera (CFOV) to CHS. The performance of the proposed system is analyzed based on detection rate analysis. This project is developed by using MATLAB software. The analysis is done in controlled environment with 50-110 cm distance range of camera to the spool. In addition, the camera orientation is analyzed between -20º to 20º yaw angle range. In order to ensure the puller will not scratch the spool, a mathematical equation is derived to calculate the puller tolerance. By using this, the system can estimate the spool position based on the camera orientation and distance calculation. Application of this system is simple and cost-effective. A Modified Circular Hough Transform (MCHT) method is proposed and tested with existing method which is Circular Hough Transform (CHT) method to eliminate false circles and outliers. The results of the analysis showed detection success rate of 96% compared to the CHT method. The proposed system is able to calculate the distance and camera orientation based on spool image condition with low error rate. Hence, it solves the peg-in-hole problem without using Force/Torque sensor. In conclusion, a total of 7 analysis consist of image preprocessing, image segmentation, object classification, comparison between CHT and MCHT, illumination measurement, distance calculation and yaw angle analysis were experimentally tested including the comparison with the existing method. The proposed system was able to achieve all the objectives.