AUV的图优化海底地形同步定位与建图方法

海底地形辅助导航技术可为智能水下机器人(AUV)的长时间水下作业提供有界精确导航结果,且无需声学基阵辅助。然而,地形辅助导航需要已知高精度先验海图,这大大限制了其应用范围。为解决海底地形未知条件下的海底地形辅助导航问题,提出了海底地形同步定位与建图技术。利用水深预测手段和地形匹配方法进行帧间数据关联构建和闭环检测,并通过图优化技术进行海底地形同步定位与建图问题求解。仿真试验结果显示,在航行1h后,海底地形同步定位与建图系统的建图结果中所有测深点定位误差均值约为13m。

AUV Bathymetric Simultaneous Localization and Mapping Based on Graph Optimization Method

Bathymetric terrain aided navigation (TAN) could provide bounded accurate navigation results for long-term underwater operation of autonomous underwater vehicles (AUVs) without assistance of acoustic arrays. However, the requirement of a high-resolution prior map confines the application of TAN. To deal with the problem of applying TAN in unknown environment, the bathymetric simultaneous localization and mapping (SLAM) algorithm is proposed. Frame-to-frame data associations and loop closures are detected via depth estimated method and terrain matching process, respectively, and the bathymetric SLAM problem is solved by graph optimization method. According to the simulation experiment results, after the AUV sailing for an hour, the average location error of all measured points in the constructed map is approximately 13m.