A fuzzy controller prototyping shell, and its application in a sonar-sensing mobile robot control system

This thesis presents the results of the investigation, implementation and testing of a tool to support rapid-prototyping development of high performance fuzzy controllers for mobile robotics. Fuzzy inference is a strong candidate for the development of robot control systems because it represents the continuous real-time inputs and outputs necessary for control as convenient textual rules. This thesis shows how the inflexibility of traditional fuzzy inference development methods can be overcome by using a free-form rule-base similar to an expert system. A rapid-prototyping approach can then be taken to controller development.