Fuzzy control to simulate 4 autonomous navigation behaviors in a differential-drive mobile robot

This paper comprises the algorithmic design and simulation of an autonomous navigation system for a differentialdrive mobile robot composed of 4 Mamdami-based fuzzy controllers. To do so, Matlab and the mobile robots simulator Sim.I.am were used. For the control design, each of the behaviors or controllers were modularly designed and tested following a reactive architecture. Subsequently, they were merged into a navigation system by the design of a finite-state machine, which selects the most appropriate behavior the robot must follow at each time step. To test the approach, several simulations were carried out for each modular fuzzy controller and for the integrated system. As a consequence, the potential of fuzzy logic in conjunction with the finite-state machine coordination yielded smooth paths, short navigation paths and few state changes, providing robustness to high-frequency state changes.