Design and Implementation of an Autonomous Vehicle to Collect Tennis Balls Using Artificial Vision

Abstract

The objective of this work was to design and implement an autonomous vehicle (robot) to collect tennis balls using different digital image processing techniques. The robot was built from an Arduino Nano microcontroller. A radio frequency antenna NRF24L01 receives the data from the control stage and the locomotion system integrated by motors and an odometry system composed of MPU6050 gyroscope encoders; additionally, the system has an emitter module that consists of an Arduino Uno and an antenna with the same characteristics. The prototype consists of two separate subsystems, one for collecting and processing information and the other specific for the vehicle on the ground. It is equipped with a Kinect camera that captures information from a defined area for image processing through a visual control algorithm that detects the balls by color and shape segmentation, determining their location in rectangular coordinates and sending them to the robot through a data transmission system. The Ackerman configuration mobile robot equipped with the wireless communication system receives the coordinates to carry out the movements that are controlled by sensors located on the wheels, with a maximum capacity of 4 balls. The complete running of the system obtained an accuracy of 96.9% in the collection of balls; it should be noted that the tests were carried out with several distractors whose objective was to confuse the system; these tests were carried out at various times the day in a real scenario.