CNN-based localization robot system

Abstract

This project aims to develop a robot learning system that combines a Convolutional Neural Network (CNN) and Universal Robots UR5e robotic arm for the task of picking up oranges. Leveraging the capabilities of a RealSense Camera mounted on the robotic arm, the robot system predicts the coordinates of the orange, enabling efficient and accurate manipulation. The Convolutional Neural Network is trained to recognize and locate the orange within the camera’s point of view, providing real-time data to guide the robotic arm’s movements. Combining advanced machine learning models with robotics, the robot system offers a versatile solution for automated fruit picking, with potential applications in agricultural automation and warehouse logistics. This project highlights the potential for enhancing efficiency and productivity in various industries.