Vision-Guided Robotic System for Automated Waste Sorting

This research paper aims to tackle the waste sorting problem, traditionally managed through sensor-based technologies or manual labor. In this paper, we propose an automated waste sorting solution using a vision-guided robotic system, integrating a six-degree-of-freedom (6-DOF) ABB Gofa 15000 robotic arm with a computer vision-guided system. The system employs the YOLOv11n-seg object detection model for real-time waste classification and localization, achieving a mean Average Precision (mAP50-95) of 0.893. Additionally, a hand-eye calibration method using checkerboard patterns and ArUco markers significantly improved robotic positioning accuracy, reducing positional error from ±15.4 mm to ±5.2 mm. The system successfully performed automated pick-and-place operations with a 95% success rate, demonstrating its potential for enhancing efficiency and accuracy in industrial waste sorting applications.