Agents Based Control System Design For Humanoid Robot

Humanoid robot has many joints, such as arms, neck, waist and legs. To combine all these joints together and coordinate them to accomplish desired tasks, what the humanoid robot needs is a control system that has high real-time ability and stability. Therefore, to choose a suitable control system is significant to a humanoid robot system.To improve humanoid robot stability and real-time performance in unexpected conditions, this paper presents a novel hierarchical control structure based on multi-agents, including main control layer, communication layer and execution layer. The execution layer consists of some agents and each agent has its own controller, motor drivers and sensors. Each agent not only can detect self-collisions between robot links cooperating with main control layer, but also has the ability to detect and deal with some unexpected conditions of the robot directly. Thus the computation burden on main control layer is reduced and real-time performance of the system is improved.To realize the control performance of the control system, I employ the processor of TMS320F2812 of TI Corporation and CPLD of Xilinx Corporation to design the joint controller, which has high computation and communication performance. Besides, I choose the real-time kernel operating system uC/OS-II as the operating system of the controller,...