Intelligent object sorting system using Dobot Magician and computer vision

Abstract

This study aims to analyze the relationship between object color, shape, and count and the performance of a robotic object sorting system, measured by execution time and accuracy. The Dobot Magician robot was used for object manipulation, while computer vision based on object detection was implemented using Python and an open source computer vision library (OpenCV). Color and shape segmentation were performed using the hue, saturation, and value (HSV) color space and OpenCV library. Experimental results indicate that three shapes (circle, triangle, and square) and four colors (green, red, blue, and yellow) affect picking time and accuracy differently. Squares generally required the longest time to pick, while triangles often had the shortest times across various colors. Accuracy remained consistently high across all colors and shapes, with green and yellow objects showing slightly higher average accuracy. These findings provide valuable insights into the factors influencing the performance of automated robotic grasping systems, which can be applied to optimize robot design and improve the efficiency and precision of object sorting tasks.