【作者】 薛颂东； 曾建潮； 杜静；

【Author】 Xue Songdong~(1,2),Zeng Jianchao~2,Du Jing~3 1.Col.of Elect.& Informat.Engn.,Lanzhou University of Technology,Lanzhou 730050,China 2.Complex Syst.& Computational Int.Lab,Taiyuan University of Science and Technology,Taiyuan 030024,China 3.North Automatic Control Technology Institute,China North Industries Group Corporation,Taiyuan 030006,China

【机构】 兰州理工大学电气工程与信息工程学院； 太原科技大学复杂系统与计算智能实验室； 中国兵器工业集团公司北方自动控制技术研究所；

【摘要】 以目标搜索定位任务为背景,用扩展微粒群算法对具有非完整运动约束特性的自主移动轮式机器人组成的群机器人系统实施协调控制.据机器人与微粒的物理属性及搜索行为等方面的特征异同,提出群机器人搜索与微粒群算法的映射概念.在定义机器人的邻域结构及时变特征群基础上,用扩展的微粒群算法对群机器人系统抽象建模.散布在搜索空间中的机器人并发检测目标发出的多源异类信号,各自融合后交互比较以确定自身最优认知和特征群的社会最优位置.个体机器人将自身惯性和经验及特征群的社会经验综合后根据群体智能原则迭代计算得到机器人在各控制时刻的平动线速度和期望位置,并由此得到转动角速度,进而按机器人的运动学特性转化为包含平移线速度和旋转角速度的控制向量,作为个体控制器的输入对机器人实施控制.该分布式算法并发运行于各机器人的板上处理器.仿真结果表明了其有效性.更多还原

【Abstract】 Taking target search with swarm robots for instance,we explore an approach to control swarm whose members are autonomous wheeled mobile robots with non-holonomic constraints.Comparing the differences and similarities between robot and particle in properties and behaviors,the authors map swarm search to particle swarm optimization (PSO).Given definitions of neighborhood structure and time-varying character swarm of robot,we extend PSO to model swarm robotic system at an abstract level.Multi-source heterogeneous signals of target are detected and fused independently by each robot in parallel,being used to decide the best-found positions both of robot itself and of character swarm by comparison.Then the expected positional series can be gained in iteration control by synthesizing its inertia and experience as well as experience of its character swarm.Finally,control vector consisting of linear and angular velocity is translated as available control inputs to individual controller at each time step,depending on robot kinematics.By this way,swarm robots can work together cooperatively.Simulation results indicate the validity of our control strategy and designed algorithm.更多还原