机器人关节通信总线系统的建模与验证

摘要

高速串行现场总线(controller area network,简称CA)被广泛部署到机器人通信系统中.而服务机器人任务具有并发性和高实时性的特点,因此,如何根据总线协议规范和应用需求精化设计模型,保证系统设计的正确性和实时性要求,避免设计阶段的漏洞十分必要.针对传统方法的局限性,提出使用形式化方法对基于CAN现场总线型控制系统进行建模分析.首先,对系统进行模型抽象和形式表达;其次进行形式建模和自动验证,在UPPAAL中实现主控制器、关节控制器、收发器、仲裁器和CAN总线的时间自动机模型;最后对机器人通信系统进行正确性验证和实时性分析.实时性分析发现:随着总线上关节节点数的增多,低优先级节点的最坏仲裁时延的增长速率加大.针对这个问题,在形式模型中加入了改进的动态优先级策略.实验结果表明:部署动态优先级策略后不仅减小了低优先级节点的仲裁时延,而且还可以加大CAN总线的节点负载量,为系统设计提供有效的指导和参考.%Controller area network (CAN) is a high-speed serial fieldbus,widely deployed in the robot communication system.Due to the concurrency of service-oriented robot tasks and tightness requirement for real-time,it is necessary to explore how to refme the design model according to the bus protocol specifications and application system,in order to ensure the correctness and real-time requirements of the system and avoid bugs in design process.However the traditional methods are limited.This paper proposes a formal method to verify and analyze the correctness of CAN based fieldbus real-time control system.The model abstraction,formal modeling and automatic verification are presented for the system including the time automata model of master,joint controller,transceiver,arbitration and CAN bus.The formal models show that the worst arbitration delay of the low priority node increases rabidly with the increasing number of joints on the CAN bus.Furthermore,an improved dynamic priority strategy is designed and added into the formal models in order to improve the worst arbitration delay problem.The experimental results show that the deployment of the dynamic priority strategy not only reduces the arbitration delay of the low priority nodes,but also increases the capacity of nodes on CAN bus.The result provides guidance and effective reference for the system design.