Implementation of Object Detection by Color in HSV Space on Robot Arm

Abstract

This research presents a colour-based object detection system utilizing the HSV colour space and a decision tree model. Achieving an overall precision, recall, and F1-score of 0.98, the system demonstrates robust detection capabilities across various cube colours: blue (0.87-0.89), green (0.87-0.92), yellow (0.74-0.80), red (0.77-0.89), and purple (0.89-0.94). Additionally, the performance evaluation of a robot arm in object manipulation reveals an 80% success rate and a 20% failure rate. Testing across five trials per colour shows successful identification and manipulation distributions for blue (4 successes, 1 failure), green (4 successes, 1 failure), yellow (4 successes, 1 failure), red (4 successes, 1 failure), and purple (5 successes, 0 failures). These findings suggest potential applications in industrial automation tasks requiring reliable colour-based object classification and manipulation, albeit with opportunities for further enhancement to minimize failure rates.