Adaptive neuro-fuzzy control of AMRU5, a walking robot with 6 pantograph-based legs

In this paper, we show an original method to derive initial parameters of Takagi-Sugeno Fuzzy Logic Controllers based from Ziegler-Nichols methods. These controllers are designed on each joint of legs of the AMRU5 robot. They will work properly in a closed operational range. Then these controllers are made adaptive to take into account the change of the payload of the robot, the change of its environment, the uncertainties etc. To do that four original steps have been designed that are: the identification of the dynamic model of the legs joints, the estimation of the parameters of the dynamic model of the legs, the calculation of an ideal torque from the estimated parameters. This torque is used in the updating of the parameters of the controller. And finally, the design of the control terms to assure that the errors during the tracking trajectories of the joints of the legs are bounded.