基于激光传感器室内机器人自主导航技术研究

发布时间：2018-05-19 01:31

本文选题：室内机器人 + 激光传感器　；参考：《天津理工大学》2017年硕士论文

【摘要】：目前定位及避障方法存在技术实现困难、实时性差、定位误差较大,系统维护成本高等问题。有鉴于此,本文综合系统运行的稳定性、实时性、可行性、正确性等要求,旨在对移动机器人定位与路径规划方法上进行研究并改良。具体研究内容如下:针对航迹推算法存在的轨迹累计误差和传统扩展卡尔曼滤波算法定位精度较差的问题,结合已知的结构化环境,本文给出了一种改进的基于扩展卡尔曼滤波器结合地图匹配方法的定位方法。该方法从传感器信息中快速提取边界角点信息,建立机器人运动模型及控制输入模型描述关系。根据特征点在世界坐标系与局部坐标系的映射关系,通过将构建的预测模型与已建模的结构化环境信息进行匹配确定机器人在目标场景中的位姿空间状态。路径规划要求机器人在运行过程中能够识别障碍物并且以最小或者较小的代价避开障碍物完成自己的任务。由于室内机器人工作场景地面路况较为平整且结构较为简单,因此本文只考虑水平面层次上的路径规划,我们将激光传感器感知的障碍物简化构造为圆柱形,其水平截面为圆形。然后通过提前设定的阈值对激光雷达采集的数据信息聚类分析,识别障碍物的数据点集合并运用最小包围圆算法构建障碍物特征模型,然后通过将整个路径规划过程分解为数个子问题进行求解,并将最小偏转角作为驱动力解决机器人运行过程中避障与寻径问题。
[Abstract]:At present, the localization and obstacle avoidance methods are difficult to achieve, poor real-time, large positioning error, high system maintenance cost and so on. In view of this, this paper synthesizes the requirements of stability, real-time, feasibility and correctness of the system, aiming to study and improve the method of location and path planning of mobile robot. The specific research contents are as follows: aiming at the problem of track accumulative error and poor positioning accuracy of traditional extended Kalman filter algorithm, combined with the known structured environment, In this paper, an improved location method based on extended Kalman filter and map matching is presented. In this method, the boundary corner information is quickly extracted from the sensor information, and the robot motion model and the control input model are described. According to the mapping relationship between the feature points in the world coordinate system and the local coordinate system, the position and pose space state of the robot in the target scene is determined by matching the constructed prediction model with the structured environment information. Path planning requires the robot to identify obstacles and to avoid obstacles at a minimum or a small cost to complete its task. Because of the flat road condition and simple structure of indoor robot working scene, this paper only considers the path planning at horizontal level, and we simplify the barrier of laser sensor perception to cylinder shape. The horizontal section is circular. Then the data collected by lidar are clustered and analyzed by the threshold set in advance to identify the set of data points of obstacles and to construct the obstacle feature model by using the minimum bounding circle algorithm. Then the whole path planning process is decomposed into several sub-problems, and the minimum deflection angle is used as the driving force to solve the obstacle avoidance and path seeking problem in the process of robot operation.
【学位授予单位】：天津理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP242

【相似文献】