【作者】 覃柯；

【导师】 孙昌志； 孙茂相；

【作者基本信息】 沈阳工业大学 ， 控制理论与控制工程， 2004， 硕士

【摘要】 移动机器人是机器人家族中的一个重要分支，也是进一步扩展机器人应用领域的重要发展方向，因此对移动机器人系统的研究，一直都受到学者们的普遍关注。本文针对多移动机器人的协作及其运动规划的问题进行了研究。通过对一个具体问题进行分析，将多移动机器人的协作及其运动规划的问题分解为行为决策与路径规划两个层次，并分别对这两个方面进行了深入的探讨和研究。主要研究成果如下： 1、充分研究了多移动机器人的协作方式及在运动态境中的实时规划中存在的主要问题及现有的各种解决方法。并针对一个具体的问题讨论如何利用基于分层规划的方法来解决复杂的协作及运动规划的问题。 2、对于行为决策层，利用分段的递推最小二乘辨识方法结合线性模型对障碍物下一步的运动进行准确预测，为高效的多机器人协作打下了基础。使用一个混合选择器即传统的产生式系统与BP网络相结合的动作选择器来进行动作选择及角色分配上，混合动作选择器在一定程度上达到了产生式系统和BP网络的优势互补，因而取得比较令人满意的效果。 3、对于运动规划层，提出了一种用于动态环境下的改进的人工势场法。该方法主要从两方面对传统人工势场法进行了改进：使用基于势场强度方法来取代矢量合成方法，并且通过引入一个填平势来引导机器人走出局部极小点，克服了传统人工势场法的缺陷；将障碍物的运动信息引入到排斥势函数中，使传统的静态势场变为动态势场。通过这种方法可以使机器人对于移动障碍物的避碰更有效。 仿真实验证明了本文所提规划方法的正确性和可行性。更多还原

【Abstract】 As a important part in the Robotic, the study of the mobile robot is also the focus of the development of the robot application. In this thesis, the cooperation and the motion planning of multi-robot system is studied. Through the analysis of a real case, the problem of the cooperation and the motion planning of multi-robot system has been divided into two parts: the upper level is action-selection and the elementary level is motion planning.. The main results of these studies are followings:1. All the questions concerning the cooperation ways of the multiple mobile robots and the main problem about real-time motion planning in the dynamic environment has been fully studied. Base on a practice case, a kind of stratified planning method has been proposed to deal with the problems of the complicated multiple robot system.2. In the upper action-selection level, the accurate prediction about the next action has been gave out base on the segmented recursive least square algorithm, witch ensure the result of the motion planning be of high efficient. A hybrid action selector witch is a integration of BP network and production system is presented to deal with the cooperation of the multiple mobile robot system. It effectively integrates the merit of the two methods and makes the process of the planning more efficiently.3. To solve the problem of the motion planning, a kind of improved artificial potential field method applied in the dynamic environment is proposed. The method mainly improve two aspects of the traditional artificial potential field method: Using the potential field intensity-based method replace the vector confusion method; Combining the motion information into the repulsive force function, which transform the static artificial field function to the dynamic artificial field function. Using this method, moving obstacle avoidance can be performed more efficiently.Simulation show the feasibility and effectiveness of the approaches and algorithms .更多还原