一种基于点云地图的机器人室内实时重定位方法

A Method for Real-time Relocalization of Indoor Robot with Point Cloud Map

确定机器人自身在点云地图中的位置是实现机器人自主导航的关键。提出一种基于点云地图的机器人室内实时重定位方法:利用预先生成的室内机器人点云地图,通过基于词袋的地图检索与位姿图优化,确定机器人在点云地图中的初始位置;在机器人运动过程中,结合视觉跟踪与参考点云地图生成新的机器人导航地图,提高相机跟踪的鲁棒性并反应场景结构的变化,实现机器人室内导航点云地图的分发与重用。通过实验验证了方法的有效性、精确性与实时性。

It is the fundamental of auto-navigation for indoor robots to determine the location within the prebuilt point cloud map.Proposed a method for real-time relocalization for indoor robots based on prebuilt point cloud map:determined the initial position of indoor robots within prebuilt point cloud map with the help of Bo W and pose graph optimization;during the movement of indoor robots,created new point cloud map through combination of visual tracking and referenced point cloud map to improve robust of camera tracking and detect the change of scene structure,implemented the distribution and reuse of point cloud map for indoor auto-navigation of robots and verified the effectiveness,accuracy and real-time through experiments.