Main Article Content

Abstract

This research enhances the performance of a wheeled soccer robot in the RoboCup Middle Size League (KRSBI-B) by integrating Stereo Vision cameras, YOLOv5, and Harris Corner Detection for precise distance estimation and object detection. The objective is to improve the robot's ability to accurately recognize and measure the distance of objects, particularly the ball and opposing robots. Using image processing, the system significantly enhances real-time object detection, improving decision-making during the match. The YOLOv5 algorithm, trained with 4,000 labeled images, achieved impressive accuracy with confidence levels up to 0.99 for ball detection at 250 cm. Results show strong correlations between high-confidence detections and accurate distance estimations, enabling effective responses to dynamic match situations. This system provides a competitive edge, improving responsiveness, adaptability, and gameplay strategies, supporting its application in real-world robotic competitions.

Article Details

References

Read More