Autonomous control of a humanoid soccer robot : development of tools and strategies using colour vision : a thesis presented in partial fulfilment of the requirements for the degree of Master of Engineering in Mechatronics at Massey University

Abstract

Humanoid robots research has been an ongoing area of development for researchers due to the benefits that humanoid robots present, whether for entertainment or industrial purposes because of their ability to move around in a human environment, mimic human movement and being aesthetically pleasing. The RoboCup is a competition designed to further the development of robotics, with the humanoid league being the forefront of the competition. A design for the robot platform to compete at an international level in the RoboCup competition will be developed. Along with the platform, tools are created to allow the robot to function autonomously, effectively and efficiently in this environment, primarily using colour vision as its main sensory input. By using a 'point and follow' approach to the robot control a simplistic A.I. was formed which enables the robot to complete the basic functionality of a striker of the ball. Mathematical models are then presented for the comparison of stereoscopic versus monoscopic vision, with the expansion on why monoscopic vision was chosen, due to the environment of the competition being known. A monoscopic depth perception mathematical model and algorithm is then developed, along with a ball trajectory algorithm to allow the robot to calculate a moving balls trajectory and react according to its motion path. Finally through analysis of the implementation of the constructed tools for the chosen platform, details on their effectiveness and their drawbacks are discussed.