Fruit Harvesting Robot Using Computer Vision

Abstract

Agriculture has conventionally been a labor-intensive occupation in India. However, in order to provide for the rapidly increasing population, in the face of rising labour costs, there is a need to explore autonomous alternatives in place of traditional methods. This paper proposes a prototype of an autonomous fruit harvesting robot consisting of a robotic arm erected on a mobile chassis. Our proposed design is capable of identifying fruits with the help of a camera module using image preprocessing supplemented with object detection algorithm (YOLO v3). We also qualitatively compared two models, one based only on image processing and the other based solely on object detection algorithm (YOLO), while taking into account the shape and colour of the fruits. When fruits are recognized, the robotic arm is engaged, and the fruit is picked and stored in the container attached to the robot's body. To pick and arrange the fruits, an end effector subsystem is used. We have also used sensors to collect important data like humidity, temperature, and rain for further processing.