基于Multi-Agent的嵌入式六足机器人系统设计

Design Method of Embedded Hexapod Robot System Based on Multi-Agent

针对目前集中结构和多层式架构控制系统实时性较低、容错能力较弱等问题,根据六足机器人系统对控制性能的要求,提出一种将多智体概念应用于嵌入式六足机器人控制系统的设计方法。该方法利用多智体相关概念分析、求解分布式多处理器机器人控制问题。与传统体系结构相比,基于MAS建立的分布式控制结构机器人系统有容错性强、集成能力强、可扩展性强等优点。智能体以嵌入式微控制器ARM为硬件核心,以μC/OS-Ⅱ和Linux操作系统为软件平台,各智能体通过CAN总线实现互连和信息交互,共同协调完成控制任务,实现无线控制与视频传输功能。经实验证明,该设计可以有效提高机器人控制性能,并且具有可移植性、可靠性及实时性强等优点,为多处理器复杂控制系统提供了一种将系统功能定义到具体实现的有效解决方案。

According to the requirements of six-legs robot system for control performance,we present a new design to solve the problem that traditional centralized architecture and multi-layer robot architecture are poor in real-time performance and weak fault tolerance.The new design is based on multi-agent system(MAS)concept,which uses the multi-agent concept to analyze and solve the distributed multi-processor robot control problem.Compared with the traditional centralized architecture and multi-layer robot architecture,the MAS-based distributed control structure has strong fault tolerance,integrated ability,scalability and other advantages.The agent takes the micro-controller ARM as the hardware core,the embedded real-time operating systemμC/OS-Ⅱand Linux as the software platform.The agents can realize the interconnection and information interaction through the CAN bus,coordinate and complete the control tasks,and realize the wireless control and video transmission function.Experimental results prove that the design can improve the robot control performance effectively,and has the advantages of portability,reliability and real-time.The paper also provides an effective solution for the design of complex embedded controllers from the definition of system functions to the realization of platform construction.