Automation of plastic waste sorting through robotic technology

Chong, Yoong Kiat (2025) Automation of plastic waste sorting through robotic technology. Final Year Project, UTAR.

PDF Download (5Mb)

Abstract

This project presents the design, development, fabrication and evaluation of an automated waste sorting system integrating computer vision, robotic actuation and electronic control. The primary objective was to automate the classification and segregation of recyclable waste to improve accuracy and efficiency compared to manual sorting. The methodology involved fabricating a conveyor belt system, designing a slider mechanism to extend the Delta X robotic arm’s reach and equipping the robot with a vacuum gripper for pick-and-place operations. A YOLOv8 deep learning model, trained on a custom dataset of waste images, was integrated with the ByteTrack algorithm to provide real-time object detection and tracking. An ESP32 microcontroller and a Python-based GUI coordinated the conveyor, slider, robot arm and vision subsystems for seamless operation. Experimental testing demonstrated high detection accuracies of 100% for aluminium, 96% for plastics and 94% for paper. Pick-and-place success rates were 92% for aluminium, 98% for paper and 48% for plastics, the latter being affected by transparency, irregular surfaces and limitations of the IR sensor. The overall throughput achieved was 8 - 15 items per minute, with reliable continuous operation over 20 minutes, though positional drift of the robot arm and slider was observed due to the lack of feedback mechanisms. These results indicate that the prototype successfully met its objectives, demonstrating the feasibility of low-cost AI-enabled robotic sorting. Keywords: waste sorting; YOLOv8; machine vision; robotic arm; automation; object detection Subject Area: Robotics

Actions (login required)