节点文献

基于视觉和超声波传感器的地面移动机器人定位与路径规划研究

Study of Ground Mobiles Localization and Path Planning Based on Vision and Ultrasonic Sensors

【作者】 章小兵

【导师】 宋爱国;

【作者基本信息】 东南大学 , 测试计量技术与仪器, 2006, 博士

【摘要】 鉴于自主导航是地面移动机器人智能水平高低的重要标志,论文在移动机器人现有的自主导航技术基础上,对服务机器人和军警用机器人在自主定位导航时面临的关键问题进行研究,从而为移动机器人的实际应用提供理论和技术保障。设计了单超声波接收头测量多个距离的装置,该装置根据接收信号特点,系统标定时用接收信号包络峰值的0.5倍作阈值,而测量时用小阈值。采用室内环境分区法,根据空间区域特点剔除坏值,运用串联和并联推断的方式,结合该测量装置实现了机器人在动态环境下的定位。设计了一种具有方向沟道和多个位置沟道的通用视觉路标,根据路标像与像面上平均能量之间的关系,选取出合适的动态阈值进行图像分割,用扫描的办法提取出路标信息,实现了机器人在室内环境下定位。对摄像机采集来的数字图像局部图进行一阶微分后按“录取线”取阈值二值化,然后用扫描聚类的办法和成像几何关系得到机器人在楼梯上的姿态。根据机器人的运动特点,采用一种简单的控制方法实现机器人姿态调整。文中局部路径规划方法用于解决移动机器人在遥控方式下的通讯时延。在机器人前部一定范围内虚拟出八个亚目标,采集前部八个超声波传感器数据,根据工作场所障碍物分布情况,分析指定的范围内有无障碍物,结合机器人的航向偏离目标的角度,选择最佳亚目标。在运动过程中,最佳亚目标滚动更新。提出用四边形轮廓代替障碍物真实轮廓,首先进行粗略路径规划,得到一条由直线组成的最优路径,然后沿边连接得到真实路径。在起始点或目标点在四边形内部的情况下,采用短边优先来脱困。该方法具有图像预处理方便,规划出的路径精细且路径规划所用的时间少。用电磁波作为目标物的信标,根据电磁波的强弱来测距离,从而实现机器人动态寻的。该方法受天气影响小。

【Abstract】 Being an important sign of the intelligence of mobile robots, self-localization and path planning of service robots and military robots are studied in order to provide guaranties of theories and technologies for factual application of mobile robots.A single ultrasonic receiver is designed to measure multi-distance. A half of the peak of receiving signal is used as the threshold in calibration and then a small threshold is used in measure. Localization is completed in dynamic environment by dividing indoor space and eliminating bad values according to the characteristics of spatial regions and also by serial and parallel deduction.Some visual landmarks with directional channels and positional channels are designed for mobile robots to localize in an indoor environment. An appropriate dynamic threshold is selected for dividing up an image of a landmark according to the relation of energy and then the information on the landmark is picked up by scan.An appropriate threshold acting as“the lowest mark of matriculates”is used to show stairs edges obviously in a part image after one order differential of the part and then the pose information of the robot on stairs is obtained by the principle of imaging geometry and the method of scanning and clustering. A simple method of control is used to adjust the pose of the robot on the basis of the characteristic of movement.The method of local path planning proposed in the thesis is used to solve delay during communication between human operator and the remote-controlled mobile robot. The best subgoal is selected from eight virtual subgoals in a certain range in front of a robot according to the angle between the direction of the mobile robot navigation and the direction from the robot to the goal, and to the distribution of obstacles in a certain region of the measure range of eight ultrasonic sensors. During the motion of the robot, the best subgoal is refreshed according to a rolling mode.It is proposed that real outlines of obstacles are replaced by quadrangles, A path consisted of beelines is planned after coarse path planning based on quadrangles and then these beelines are connected with curve lines along borders of obstacles between them. Short side of an obstacle border is selected for robots to escape from surrounded place under the condition of the start or the goal in a quadrangle. The method is characteristic of the convenience of image pretreatment and the optimum path and shorter time spent on path planning.Electromagnetic wave is used as the signal source of a goal. The distance between a robot and the goal is measured according to the intensity of electromagnetic wave in order to search the goal dynamically. The method is affected little by environmental weather.

【关键词】 移动机器人定位导航路径规划视觉超声波
【Key words】 Mobile robotLocalizationNavigationPath planningVisionUltrasonic
  • 【网络出版投稿人】 东南大学
  • 【网络出版年期】2007年 04期
节点文献中: 

本文链接的文献网络图示:

本文的引文网络