Machine Learning Assisted Autonomous Vehicle Design and Control

Abstract

In this study, image processing is performed on the data received from the camera using the machine learning training model, and mature agricultural products are detected using image processing libraries. In addition, the location information of the product is transmitted to the motors by serial communication, and the product is collected with the help of robot arm movements, which are realized thanks to mathematical models. With the help of the developed mobile application interface, the user will be able to add new workspaces and collect vegetables autonomously on the previously saved areas. The software, which was developed to act autonomously, was originally designed using data from distance sensors and the mobile application. The user can easily ensure that the vehicle can collect one or more types of vegetables by using the mobile application interface. In this study, tests were carried out in terms of speed and accuracy by using different models for the mature product collection process with the designed machine learning and image processing based autonomous vehicle.