多机器人协作环境探测研究

发布时间：2018-03-13 03:28

本文选题：多机器人系统　切入点：协作探索　出处：《南京理工大学》2017年硕士论文　论文类型：学位论文

【摘要】：探测未知环境是移动机器人领域的基础问题,是机器人在未知环境中完成其他任务的必要条件。与单个机器人相比,多机器人系统具有更好的鲁棒性、适应性、灵活性和扩展性,更加适用于探测未知环境,但是仍然在地图融合和任务分配方面存在许多问题。地图融合直接决定了多机器系统探测未知环境的准确性,任务分配影响着多机器人系统完成环境探测的效率。本文对多机器人协作探测未知环境中的地图融合和任务分配两个重要内容进行研究。针对地图融合和任务分配中存在的技术难点,提出了解决方案,有效提高了探测效率。本文的主要工作如下:首先提出了一种几何—拓扑混合式描述的地图融合方法,将拓扑节点及其几何特征作为地图匹配的基本单元,将拓扑节点几何特征的相似度作为拓扑节点相同的判断依据,为了增加地图融合的可靠性,将拓扑节点与相邻节点的欧式距离加入到节点相似度的计算当中。提出了有效的地图融合指标和地图融合次序。其次,提出了基于市场法的多机器人任务分配策略,利用经济市场的拍卖思想,利用合同网协议,根据机器人状态随时间变化,采用二次拍卖方法进行任务分配。为了避免机器人过多的集中于同一区域,对常见的投标计算方式进行改进,利用排斥信息素进行投标计算,减少机器人对相同区域的重复探索。最后,在仿真实验平台对本文提出的地图融合算法和任务分配算法进行实验,通过设置不同的机器人数量和环境范围验证本文方法的有效性。
[Abstract]:Detection of unknown environments is a fundamental problem in the field of mobile robots and a necessary condition for robots to accomplish other tasks in unknown environments. Compared with a single robot, multi-robot systems have better robustness and adaptability. Flexibility and expansibility are more suitable for detecting unknown environments, but there are still many problems in map fusion and task assignment. Map fusion directly determines the accuracy of multi-machine systems in detecting unknown environments. Task assignment affects the efficiency of environment detection in multi-robot systems. In this paper, two important contents of map fusion and task assignment in multi-robot cooperative detection unknown environment are studied. Technical difficulties in matching, The main work of this paper is as follows: firstly, a map fusion method is proposed, in which topological nodes and their geometric features are taken as the basic unit of map matching. In order to increase the reliability of map fusion, the similarity of geometric features of topological nodes is taken as the basis for judging the same topological nodes. The Euclidean distance between topological nodes and adjacent nodes is added to the computation of node similarity. An effective map fusion index and map fusion order are proposed. Secondly, a multi-robot task allocation strategy based on market approach is proposed. Using the auction idea of economic market, using contract net protocol, according to the change of robot state with time, the second auction method is used to assign the task. In order to avoid the robot being too concentrated in the same area, The common bidding calculation method is improved, and the bidding calculation is done by using repellent pheromone to reduce the repeated exploration of the same area by the robot. Finally, The proposed map fusion algorithm and task assignment algorithm are tested on the simulation platform, and the effectiveness of the proposed method is verified by setting the number of robots and the range of environment.
【学位授予单位】：南京理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP242

【参考文献】