Fuzzy logic control vs. nonlinear P control of a three wheeled mobile robot (TWMR)

The inverse and forward kinematics modeling of a three wheeled mobile robot (TWMR) is investigated. Assuming no track-wheel slippage, the TWMR has two degrees of freedom and needs only two driving actuators/motors. The robot is modeled (inverse kinematics) and the governing equations to track a desired path are derived. To verify the results, the TWMR is modeled in forward kinematics using simulation as well as experiment. The results showed good agreement between the desired path and the trace generated by the robot. Therefore the simulated forward kinematic model of robot is used to design controllers. A specific lane change duty is defined for the robot. A fuzzy logic controller is designed and its rules are optimized using heuristics and human expertise for the best results possible. The inputs of the controller are distance from desired lane and its derivatives and the outputs are angular velocities of actuators. Finally, a nonlinear P controller is developed for the same task and is optimized using genetic algorithm. The performance of our fuzzy controller is comparable to the nonlinear P controller. However, because of simplicity and the better optimization of P controller, its results are somehow better than the fuzzy controller.