Motion Control System Design Of Humanoid Robots

Humanoid robots can be considered as nonlinear high-order systems with multiple degrees of freedom and non-complete constraints, spanning many research fields, such as mechanics, electronics, communication, computer science, bionics, etc. As a result, the research on humanoid robots may serve as an ideal experimental platform for kinetics, dynamics, simulation technology, sensor fusion, control theory and so on. As the kinematic performance of a robot is directly influenced by its motion control system, thus the kinematic research and development has remained as one key area in the research of humanoid robotics. Furthermore, the motion control system of small-sized humanoid robots has several special features. For example, the system must be capable of controlling a quantity of joints within a finite space, responsible for the joint coordination as well as guaranteeing real-time performance. In addition, the motion control system also poses more restrictions for the dimension and weight of the system layout.This paper presents an improved motion control system structure for humanoid robots, by incorporating the advantages of both centralized and distributed control systems. Such a structure is made up of coordination motion module, data communication module and the motion control modules distributed on the bottom layer. The bottom control units are responsible for the control of robot arms and legs of, each of which includes joint controller, joint drive, and a number of sensors. The structure takes the actual robot configuration into consideration, which is that those adjacent joints usually share similar functionalities and thus form a tight coupling relationship in-between. By this means, the control system will not only preserve the intelligence and flexibility from distributed structure, but greatly reduce data communication load, so that it will better serve the needs for robot control and sensory.The hardware system is then constructed based on high-performance Digital Signal Processor, specialized motor control module, power drive module and Quadrature Encoder Pulse module. The hardware platform, along with the software program and control algorithm design, fully exploits the potentials of DSP chips, and enables the simultaneous control of multiple motors and comprehensive information sharing based on a variety of communication buses. The ultimate integrated system is more universal, more compact, and easier for coupled control between robot joints.Experimental results carried out on MIH-I Humanoid Robot platform demonstrate that such a system design is suitable for actual motion behavior requirements and achieves satisfactory control precision.