Reactive Navigation of Autonomous Vehicle using Neuro-Fuzzy Controller

This paper describes neuro-fuzzy logic controller approach to reactive navigation of an autonomous vehicle. This work also describes an unconventional approach to decision making and defuzzification when the output of the controller is discrete in nature. The decision making is done by voting and defuzzification reduces to the use of a simple max operator, in other words a 'winner takes all' approach. This work is based on experiments conducted on a small autonomous vehicle, which runs on stepper motors.