Development of an automatic fruit picking robot based on object detection algorithm

• Main Author: Wan, Jiaxin
• Other Authors: CHEAH Chien Chern
• Format: Thesis-Master by Coursework
• Language: English
• Published: Nanyang Technological University 2020
• Subjects: Engineering::Electrical and electronic engineering::Control and instrumentation::Robotics
• Online Access: https://hdl.handle.net/10356/142532
• Institution: Nanyang Technological University

Robots are increasingly involved in human production, which leads to a qualitative leap in production efficiency. In the agricultural field, especially in the process of harvest, it needs a lot of human resources to perform various tasks, such as picking fruits, which may be labor intensive. An effective robotics picking system can solve this problem to a large extent. This dissertation aims to develop a robot system for automatic picking based on a real-time CNN (Convolutional Neural Network) algorithm, that is, YOLO (You Only Look Once). In this project version 3 of YOLO is used based on Keras to detect ripe fruits(kumquat) and RGB-D camera is used to obtain the physical coordinates of the fruits. After locating the objects, an industrial manipulator UR5e robot is programmed to pluck the fruits automatically.