Humanoid Robot Based On The Embedded System Platform Balance And Movement Control

Robots, which combine the newest scientific research achievements in such subjects as mechanical engineering, electronic engineering, computer engineering, automatic control engineering, artificial intelligence and so on, represent the highest achievement of mechatronics technique. Nowadays, research on humanoid robot has become one of the most active research areas in robotics. Among them, balance control technology and motion control technology are important components in the area of humanoid robot. Because of large number of joints and complicated structures, its control strategy and control circuit would be very complex, if the humanoid robot functions like human beings. Researches on balance control and motion control for humanoid robot are conducted deeply and systemly in this paper.Focusing on the balance control problem of walking for humanoid robot, a fuzzy b-spline neural network controller (FBSNNC) is proposed and applied to the balance control of humanoid robot. Stable walking control of humanoid robot is implemented by adjusting the forward and backward swing angles of ankle joint and making the actual Zero Moment Point(ZMP) coincident with the desired ZMP. Simulation results show that FBSNNC allows control system to have a fast response speed as well as good steady-state precision, in addition, system oscillation is restrained strongly. Feasibility and efficiency of above FBSNNC in balance control of humanoid robot are confirmed.S3C44BOX processor, which is a product of SUMSUNG Electronic Corporation, is chosen to establish a motion control platform for humanoid robot. The motion control platform can control numbers of motors simultaneously. It possesses human-computer interface and can also response abnormal motion status via external interrupt.CAN-bus is utilized to realize the real-time communication between each control unit in this paper. Hardware circuit design, program design, and corresponding experiment research work are carried out for motion control and CAN-bus node. Experiment results indicate that the above motion control platform can implement multi-joint motion control as well as data communication among control units. In order to ensure control tasks to be completed accurately, a embedded real-time operation system is necessary to schedule multi-task for the control system of humanoid robot.μC/OS-Ⅱembedded real-time operation system is chosen and transplanted into the embedded processor, three tasks are established to simulate multi-task scheduling experiment of operation system. Experiment results show that the embedded real-time operation system can realize multi-task scheduling, and implement cooperative motion control for the joints of humanoid robot.