To address the issue of low positioning accuracy of traditional sonar inertial systems in weakly textured un-derwater environments,it is proposed to apply sonar inertial filtering odometry to underwater sonar inertial system position-ing.Based on the principle of sonar imaging,feature extraction is performed on sonar images,and iterative nearest point al-gorithm(ICP)is used for feature matching.By fusing the point cloud matching results with IMU data,the stability of state estimation is improved.This method uses IMU data for state prediction and updates the state based on sonar ICP matching results as observations.Finally,the filtered odometer and ICP sequential matching results will be incorporated into the graph optimization framework for pose optimization.The lake experiment shows that compared to traditional methods,this meth-od has higher positioning accuracy and can adapt to weak texture environments.
关键词
声呐惯性定位/滤波里程计/迭代最近点算法/点云匹配
Key words
sonar inertial positioning/filter odometer/iterative nearest point algorithm/point cloud matching
维普
万方数据