Neuro-fuzzy based system for mobile robot navigation.

The main topic of this thesis is the neuro-fuzzy control approach for mobile robot navigation in indoor environments. A system for directing the robot towards any target, point has been developed and tested in simulated environments. The autonomous mobile robot uses only information obtained from its infrared and contact sensors in order to reach the target point and to avoid collisions. The control system is organized in a top-bottom hierarchy of various tasks to be executed by the robot, tasks which correspond to specific robot behaviors during his navigation. The task to be executed at every control cycle is determined according to the robot behavior required to be performed at that time. Each task is implemented through fuzzy (or neuro-fuzzy) techniques. When multiple low-level behaviors are required, a command fusion method is used to combine the output of several fuzzy sub-systems. A switching coordination method is used to select a behavior among higher level behaviors