【作者】 杨涛；

【导师】 黄心汉；

【作者基本信息】 华中科技大学 ， 控制理论与控制工程， 2005， 硕士

【摘要】 机器人技术的发展使得机器人的能力不断提高,其应用的领域和范围正不断扩展。人们希望机器人能完成由单一机器人难以完成的更加复杂的作业。这些复杂的作业,需要多个机器人相互协调与协作共同完成,因此对多机器人协调与群体协作的研究就显得非常重要。目前,关于多机器人系统的研究正成为一个新的研究方向——协作机器人学。本文结合机器人足球比赛和移动机器人编队,基于分布式人工智能研究领域多智能体的概念,对移动机器人控制和多机器人协作问题进行了研究,提出了一些新的方法。本文探讨了多智能体系统的组织和协调,提出了一种基于MAS 的多机器人系统模型,这种分层结构模型有效地把任务的决策、规划、信息的感知和实时执行集成起来,将系统功能和实体机器人行为统一起来,从而使系统的程序空间与它所描述的现实空间有较一致的对应关系,适合于多机器人系统的分析与建模。移动机器人运动控制是建立在运动模型基础上的。通过对轮式移动机器人的物理结构进行分析,得到移动机器人的运动模型。在该模型的基础上,对机器人到某一点的轮速算法,转角的比例轮速算法和直线精确到点算法这三种基本运动规划进行了分析。在此基础上,对机器人复杂行为规划进行了讨论。足球机器人比赛是典型的多Agent 系统,在机器人足球比赛策略流程的基础上,对决策系统进行了分层次的分析,进行了决策系统中球队攻防转换策略和角色分配策略的详细设计,并且着重研究了基于神经网络的信息融合方法在机器人足球决策中的应用。针对机器人任务分配以及分配需要遵循的原则进行了量化分析,对多机器人团体协作的高层策略进行了研究。在Pioneer2移动机器人平台软硬件基础上构建了一个基于Windows 的移动机器人通信平台,给出了该平台的具体设计和实现,并且对移动机器人编队控制进行了初步研究。机器人足球比赛成绩和多机器人编队的实验结果及分析证明了上述方法的有效性和正确性。更多还原

【Abstract】 With the development in robotics, the capability of robot is improved ceaselessly and its application field has been extended. Robots are being expected to do more complex task which a single robot has no ability to accomplish. Obviously, these tasks need coordination and cooperation among multiple robots. So it is important to study multi-robot cooperation. Recently, the research of multi-robotic systems is becoming a new research direction—Cooperative Robotics. Based on the concept of multi-agent systems in DAI, robot soccer game and mobile robot formation, the dissertation gives a research on mobile robot control and multi-robot cooperation and puts forward some new methods. The dissertation discusses the organization and coordination of MAS and puts forwards a kind of multi-robot system model based on MAS. It integrates effectively decision of tasks, planning, information and real-time execution. It can also unify the function of system and robots in order to analyze and model multi-robot system. The model of motion is the basis of mobile robot motion control. According to the analysis of the physical structure of wheeled mobile robot, motion model can be constructed. Under the basis of it, the dissertation analyzes three velocity algorithms: reaching the fixed point, rotating and reaching points accurately. Then the dissertation discusses the complex action planning. Robot soccer game is a typical multi-agent system. The dissertation analyzes the decision system separately according to the flows of robot soccer game. The design of the switch of attack and defense and the strategy of role assignment are put forwards. And the method of information fusion based on neural networks is researched under the background of robot soccer decision. The dissertation also analyzes quantificationally the tasks assignment and the principles, and studies the high level strategy of multi-robot cooperation Moreover, the dissertation constructs a Windows communication platform based on software and hardware of Pioneer2 mobile robot. The content realization and the protocol format of data package are given. The formation control is also studied. The achievements of robot soccer competition and the multi-robot formation experiment’s results verify that the above methods are correct and effective.更多还原