融合ArUco码先验信息的SLAM累计误差消除算法

A Method for Eliminating SLAM Cumulative Error Fusing Prior Markers

移动机器人在无回环场景通过双目视觉传感器估计位姿时,存在累积误差无法消除的问题。为解决此问题,提出了一套利用ArUco码先验信息辅助消除累积误差的算法。该算法由ArUco码检测、相机位姿估计和累积误差消除三部分组成。首先,通过阈值化、四边形检测及亚像素优化等图像处理方式检测到ArUco码在图像中的位置;其次,通过PNP算法及先验ArUco码信息估计当前帧相机的位姿;最后,构建携带先验ArUco码信息的图模型,以Levenberg-Marquardt算法迭代优化该图模型,消除SLAM系统的累积误差。实验结果表明,该算法能够有效地消除双目视觉传感器位姿估计的累积误差,并且满足机器人实时定位的要求。

When a mobile robot estimates its pose through binocular vision sensor in a non loop scene,the cumulative error cannot be eliminated.In order to solve this problem,an algorithm is proposed to eliminate the cumulative error by using the prior information of ArUco markers.The algorithm consists of three parts:ArUco marker detection,camera pose estimation and cumulative error elimination.Firstly,the position of ArUco marker in the image is detected by image processing methods such as thresholding,quadrilateral detection and sub-pixel optimization;secondly,the pose of the camera in the current frame is estimated by prior ArUco marker information;Finally,the graph model with prior ArUco marker is constructed,and the Levenberg-Marquardt algorithm is used to iteratively optimize the graph model to eliminate the cumulative error of SLAM system.Experimental results show that this method can effectively eliminate the cumulative error of pose estimation of binocular vision sensor,and meet the requirements of real-time robot localization.