ANFIS based Controller Design for Biped Robots

This paper proposes a design method of biped locomotion controller based on ANFISs (adaptive neuro-fuzzy inference systems). In which, there is an ANFIS assigned as the system identifier to determine the model of a biped robot firstly. And then the motion controller, which is a combination of ten ANFISs, aims to perform the biped locomotion controls. One can adopt the moving velocities of biped robot, trajectory of ZMP (Zero Moment Point) and the tilt angles as the inputs of the controller, and the outputs will be the next angles of joints. As the simulation results show, the proposed controller can generate a stable walking cycle for a biped robot. Moreover, the experimental results demonstrate that the performance of proposed controller is satisfactory whenever the robot stays in different postures or moves on a rugged surface.