室内移动机器人导航系统研究与实现

Research and Implementation of Indoor Mobile Robot Navigation System

实现了一种低成本高性能室内移动机器人导航系统。针对Cartographer算法使用激光雷达数据在室内Long-Corridor场景下建图的局部匹配错误导致定位不准的问题,使用扩展卡尔曼滤波融合激光雷达、里程计和惯性测量单元3种数据进行位姿估计,得到较为精准的定位,可有效提高建图精度;针对传统AMCL算法重定位耗时长的问题,采用基于扫描匹配的重定位方法,通过将当前Scan与Submap进行匹配,降低了扫描匹配方法的重定位耗时;针对A^(*)全局规划算法路径搜索时间长、拐点较多的问题,提出一种改进A^(*)算法,通过优化启发函数和增加拐角优化函数,缩短了算法搜索时间,同时去除了冗余拐点。结果表明,重定位耗时减少80.43%,改进A^(*)算法搜索时间减少22.79%。

A low-cost and high-performance indoor mobile robot navigation system is realized.Aiming at the problem of inaccurate positioning caused by the local matching error of the Cartographer algorithm using lidar data to build maps in indoor Long-Corridor scenes,the extended Kalman filter is used to fuse the three data of lidar,odometer and inertial measurement unit for pose estimation.A more accurate positioning can be obtained,which can effectively improve the mapping accuracy;for the problem that the traditional AMCL algorithm takes a long time to relocate,the relocation method based on scan matching is adopted.By matching the current Scan with the Submap,the relocation time of the scan matching method is reduced.For the problem of long path search time and many inflection points in the A^(*)global planning algorithm,an improved A^(*)algorithm is proposed.By optimizing the heuristic function and adding the corner optimization function,the algorithm search time is shortened,and redundant inflection points are removed.The results show that the relocation time is reduced by 80.43%,and the search time of the improved A^(*)algorithm is reduced by 22.79%.