基于直线地图的机器人运动规划问题研究

发布时间：2018-05-17 22:30

本文选题：直线地图 + Voronoi图　；参考：《济南大学》2017年硕士论文

【摘要】：随着计算机科学和传感器技术的快速发展,人们对移动机器人的研究也向着智能化的方向发展。移动机器人实现智能化的基本要求是自主地运动规划,而运动规划的基础是地图构建,所以地图构建和运动规划是机器人领域重要的研究内容。移动机器人通过自身携带的传感器感知周围环境信息,构建环境地图描述环境,并在此基础上规划出能够避开障碍物的可行路径,进而在初始位置和目标位置之间搜索出一条最短路径,最后跟踪该最短路径实现从初始位置到目标位置的移动。本文主要研究了有障碍物环境下基于直线地图的机器人运动规划问题。机器人能够以激光传感器实时检测环境信息,构建直线地图描述环境,然后采用Fortune提出的Sweep Line算法生成Voronoi图,并利用Dijkstra算法在Voronoi图中搜索出从当前位置到目标位置的最短路径作为下一步运动的参考路径,最终跟踪参考路径到达目标位置。各部分前后相承,主要研究内容如下:建立了以直线段为拓扑特征的环境地图模型。利用激光传感器获取环境信息,设计了直线段生成、匹配、拟合等算法。将采集到的点的信息生成直线段,并将该直线段与全局地图中已有的直线段进行匹配,如果匹配成功,则将这两条直线段拟合生成的新的直线段导入到全局地图中;如果匹配不成功,说明该线段之前未被检测,则直接将它导入到全局地图中,实现地图的更新。基于直线地图环境模型生成描述环境信息的Voronoi图,并以Voronoi图为基础搜索生成机器人可行避障最短路径。论文研究了直线地图生成Voronoi图的方法,设计了快速生成算法;基于Voronoi图,结合Dijkstra方法设计了参考路径的搜索算法;最后编程实现了相关算法,进行了实验验证。在得到参考短路径后,论文采用Line-of-Sight(LOS)算法跟踪该路径,LOS算法将路径跟踪问题简化为角度跟踪问题。论文设计了PD控制器来跟踪LOS角度,最终控制机器人运动到目标点。基于论文设计的环境地图建模方法、Voronoi图生成算法、路径搜索算法和跟踪方法,利用C++编程实现了机器人在有障碍物环境下的运动规划功能,并在Mobile Sim软件中进行验证,实验结果证明所设计的运动规划系统是有效的、可行的。
[Abstract]:With the rapid development of computer science and sensor technology, the research of mobile robot is developing towards the direction of intelligence. The basic requirement of realizing intelligent mobile robot is autonomous motion planning, and the foundation of motion planning is map construction, so map construction and motion planning are important research contents in robot field. The mobile robot perceives the surrounding environment information through the sensors it carries, and constructs the environment map to describe the environment, and on this basis, it plans the feasible path that can avoid obstacles. Then a shortest path is searched between the initial position and the target position, and the movement from the initial position to the target position is realized by tracking the shortest path. In this paper, the robot motion planning problem based on linear map in obstacle environment is studied. The robot can use laser sensor to detect the environment information in real time, construct the line map description environment, and then use Sweep Line algorithm proposed by Fortune to generate Voronoi map. The Dijkstra algorithm is used to search the shortest path from the current position to the target position in the Voronoi diagram as the reference path for the next step and finally track the reference path to the target position. The main contents are as follows: the environmental map model with straight line segment as topological feature is established. By using laser sensor to obtain environmental information, the algorithms of line segment generation, matching, fitting and so on are designed. The information of the collected points is generated into the straight line segment, and the line segment is matched with the existing straight line segment in the global map. If the matching is successful, the new straight line segment generated by fitting the two straight line segments will be imported into the global map. If the match is not successful, the line segment is not detected before, then it is directly imported into the global map to realize the map update. Based on the linear map environment model, the Voronoi diagram describing the environment information is generated, and the shortest feasible obstacle avoidance path of the robot is generated based on the Voronoi diagram. This paper studies the method of generating Voronoi map by linear map and designs a fast algorithm. Based on Voronoi graph and Dijkstra method, the search algorithm of reference path is designed. Finally, the related algorithm is realized by programming, and the experimental results are verified. After the reference short path is obtained, the Line-of-SightSightLos) algorithm is used to track the path and Los algorithm is used to simplify the path tracking problem to the angle tracking problem. PD controller is designed to track the LOS angle and control the robot to the target point. Based on the environment map modeling method designed in this paper, the algorithm of generating Voronoi diagram, the algorithm of path searching and the method of tracking are designed. The motion planning function of robot in obstacle environment is realized by C programming, and verified in Mobile Sim software. Experimental results show that the proposed motion planning system is effective and feasible.
【学位授予单位】：济南大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP242

【参考文献】