A novel intelligent control system design for a humanoid robot

This paper presents the design of an intelligent control system for a humanoid robot. A novel fuzzy cerebellar model articulation controller (FCMAC) is proposed; this controller incorporates the fuzzy system inference rule with a CMAC fast learning ability. This FCMAC is a generalization network; in some special cases it can be reduced to a fuzzy neural network or a CMAC. This FCMAC is used as the main controller for the trajectory tracking control of the robot. In this robotic system, an inertial navigation system (INS) including gyroscopes and accelerometers is used to measure the robot´s attitude and acceleration for modifying the dynamic attitude of the robot. Moreover, a zero moment point (ZMP) compensator is used to on-line adjust the gait trajectories to improve the walking stability. The control system is implemented based on system on a programmable chip (SoPC) technology. Thus, this intelligent control system can achieve real-time on-line closed-loop feedback control of the humanoid robot. Experimental results show that the developed system can achieve favorable control performance for a high-order nonlinear humanoid robot.