A hybrid learning procedure referred to as adaptive neuro fuzzy inference system (ANFIS) is applied to an artificial leg model to generate the correct positions of the servomotors actuating the leg joints. One of the most important control problems of mechanical arms and legs is the efficient calculation of correct joint angles for a space trajectory. Although this application represents the simplest model with two degrees of freedom, the practicality of ANFIS for such mechanical systems is validated. For the gait model of the proposed mechanism, the experimental planar motion of the ankle joint is transformed to joint angles by ANFIS and approximated by polynomial functions. The corresponding servomotor positions are obtained by the proposed inverse kinematic solution method and are included in a Simulink model as an embedded Matlab function. A hybrid control system consisting of combination of a proportional plus derivative (PD) controller and a fuzzy logic controller (FLC) is applied to control the selected servomotors. The accuracy of the control system is further verified on SimMechanics.
ANFIS, inverse kinematics; Fuzzy logic; Hybrid control; Gait model