基于动态粒子蜜蜂算法的群机器人任务分配方法研究

发布时间：2018-05-07 21:13

本文选题：多机器人系统 + 任务分配　；参考：《吉林大学》2017年硕士论文

【摘要】：群机器人系统是一种典型的由多个体积小,结构简单的机器人构成的人工群体系统,对于人类和大型设备无法进入的地震和矿难等产生的危险狭小区域,由群机器人系统来工作已经是一种趋势;群机器人系统能够快速、高效地完成任务的前提就是合理的任务分配,针对不同的任务目标如何恰当分配机器人以及分配机器人的个数直接影响多机器人系统完成任务的效率,因而,受自然界中蜜蜂、鸟类等社会性群居生物通过合作能够快速觅食现象的启发,本文提出基于动态粒子蜜蜂算法的多机器人任务分配方法以提高未知环境搜索过程中的任务分配效率。首先,本文分析了同构群机器人的分层次组织结构,采用组合拍卖法建立了任务分配的数学模型,并引入访问待探测目标的顺序数,将该模型改进为一个混合整数线性规划问题,避免了解过程中产生的大量子巡回。并且,考虑到机器人对任务目标的性能值、花费时间、成本、目标优先权和机器人与目标间的距离等因素,建立了用于优化算法的适应度函数。其次,采用动态粒子蜜蜂算法优化任务分配问题,以机器人选择任务的概率最大为优化目标,结合了鸟类和蜂群的优点,利用动态粒子群算法进行蜜蜂算法的局部搜索,更新蜜蜂算法中的最佳蜂、精英蜂和侦查蜂的速度和位置,改进了蜜蜂算法容易陷入局部最优的缺陷,使优化效果更好。然后,在动态粒子蜜蜂算法优化阶段,动态粒子群进行局部搜索时加入邻域设置,保证精英蜂和最佳蜂在局部搜索过程中,更新的位置控制在位置的最大值和最小值范围内,提高完成任务的效率。最后,设计了适用于多机器人任务分配问题的仿真实验平台,依据该仿真平台进行相应实验,实验表明本文所提方法的可实现性,并且任务分配的时间比单独利用蜜蜂算法和动态粒子群算法进行分配所用的时间短;任务分配的误差比单独利用蜜蜂算法和动态粒子群算法进行分配所产生的误差小;任务分配的收敛性也比单独利用蜜蜂算法和动态粒子群算法进行分配的收敛性好,验证了基于动态粒子蜜蜂算法的群机器人任务分配方法的优越性。
[Abstract]:The swarm robot system is a typical artificial group system composed of many small and simple robots. It is a dangerous and narrow area caused by earthquakes and mine disasters that are inaccessible to human beings and large equipment. It is a trend to work by a swarm robot system. The premise that a swarm robot system can accomplish tasks quickly and efficiently is a reasonable task allocation. How to allocate robots properly and how to allocate the number of robots directly affect the efficiency of the multi-robot system to complete the task for different mission objectives, therefore, the bees in nature, In this paper, a multi-robot task allocation method based on dynamic particle bee algorithm is proposed to improve the efficiency of task allocation in the search process of unknown environment. Firstly, the hierarchical structure of isomorphism robot is analyzed, and the mathematical model of task assignment is established by combinatorial auction method, and the order number of accessing the target to be detected is introduced. The model is improved to a mixed integer linear programming problem to avoid a large number of subcircuits in the process of understanding. Furthermore, considering the performance, cost, priority and the distance between the robot and the target, the fitness function of the optimization algorithm is established, which is based on the performance of the robot, the cost, the cost, the priority of the target and the distance between the robot and the target. Secondly, the dynamic particle bee algorithm is used to optimize the task assignment problem. Taking the maximum probability of robot task selection as the optimization target, combining the advantages of birds and bees, the dynamic particle swarm optimization algorithm is used to perform local search of bee algorithm. By updating the speed and position of the best bee, elite bee and detective bee in the honeybee algorithm, the bee algorithm is easy to fall into the local optimum defect and the optimization effect is better. Then, in the optimization phase of dynamic particle bee algorithm, neighborhood setting is added to local search of dynamic particle swarm to ensure that the updated position of the elite bee and the best bee in the local search process is controlled in the range of maximum and minimum position. Improve the efficiency of accomplishing the task. Finally, a simulation experiment platform suitable for multi-robot task assignment is designed, and the corresponding experiments are carried out according to the simulation platform. The experiments show that the proposed method is feasible. The time of task allocation is shorter than that of honeybee algorithm and dynamic particle swarm optimization algorithm, and the error of task assignment is smaller than that of honeybee algorithm and dynamic particle swarm optimization algorithm alone. The convergence of task allocation is better than that of honeybee algorithm and dynamic particle swarm optimization algorithm alone, which verifies the superiority of the task allocation method based on dynamic particle bee algorithm.
【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP18;TP242

【相似文献】