Development of intelligent Municipal Solid waste Sorter for recyclables

Area of Science:

• Robotics and Automation
• Artificial Intelligence
• Environmental Engineering

Background:

• Sustainable development awareness is enhanced by Municipal Solid Waste (MSW) sorting.
• Deep learning (DL) algorithms improve MSW detection, but integration with robotic systems for factor analysis is limited.
• Intelligent MSW Sorter (IMSWS) development requires understanding key performance determinants.

Purpose of the Study:

• To develop an Intelligent MSW Sorter (IMSWS) prototype.
• To evaluate dominated factors influencing the pick-and-place sorting process.
• To integrate DL algorithms with robotic systems for enhanced waste management.

Main Methods:

• Manufactured a delta robot prototype for the IMSWS.
• Utilized a camera for RGB image and height acquisition of MSW on a conveyor belt.
• Employed YOLOv3 or YOLOv4 DL algorithms for MSW detection and location identification.
• Developed a sequence program to guide the delta robot for pick-and-place operations.

Main Results:

• The IMSWS prototype demonstrated capability in sorting multi-object MSW streams.
• Both YOLOv3 and YOLOv4 achieved high detection accuracy on the MSW image dataset.
• YOLOv4 provided acceptable detection accuracy in a moving MSW stream, though improvements are needed.
• Robotic arm gripping stability was identified as the main factor dominating IMSWS performance.

Conclusions:

• The developed IMSWS prototype shows promise for automated MSW sorting.
• Further enhancements in robotic arm gripping mechanisms are crucial for optimizing IMSWS performance.
• Integration of advanced DL algorithms and robotics offers a pathway to efficient waste management solutions.