Automatic Waste Segregation System

Abstract

This paper presents an automatic waste segregation system. The system is able to detect waste autonomously and separate it accurately by using a deep neural network for waste recognition. With waste recognition and automatic navigation functions, the system can efficiently and autonomously clean waste on the ground in places like parks or schools. The hardware architecture of the system includes a module Raspberry Pi, a webcam, PIC microcontrollers, EEPROM memory, a DC motor, a pick and place robotic arm, infrared and ultrasound sensors, a gyroscope, and GPS sensors. From the software point of view, the main parts are represented by the object detection module and a classification algorithm based on convolutional neural network architecture. This system would help in the easy recovery of recyclable items.