The task of getting a robot to move from point to point within an environment that is constantly changing without running into anything is one of the most difficult tasks to solve when designing and building a fast and robust mobile robot. This thesis records the design and implementation of a navigation system which aims to solve this problem using a behavioural approach as compared to the traditional perception model. It is shown that the behavioural model provides significant advantages in robustness, speed and reduced complexity when compared to the perception model and can also avoid many of the drawbacks involved with traditional methods of implementation such as box cannons, boundary effects and dead spots.
This thesis is one of a series which describe the components of a team of robots called the RoboRoos. These robots have been constructed by the University of Queensland, Engineering Department for entry into the RoboCup International soccer competition which tests the ability of a team of robots to coordinate together to win a game of soccer. The results and performance of the robots during the 1999 RoboCup competition show the value of the behavioral model of navigation as compared to traditional methods.