The laser gyro is most su it able for building the strap down inertial navigation system (SINS), and its acc uracy of attitude algorithm can enormously affect that of the laser SINS. This p aper develops three improved algorithmal expressions for strap down attitude ut ilizing the angular increment output by the laser gyro from the last two and cur rent updating periods according to the number of gyro samples, and analyses the algorithm error in the classical coning motion. Compared with the conventional algorithms, simulational results show that this improved algorithm has higher precision. A new way to improve the rotation vector algorithms is provided.
当前视觉同时定位与建图(Simultaneous localization and mapping,SLAM)技术已经成为导航定位领域重要的导航方式之一。为提升视觉SLAM全天候多场景的综合性能,融合多种信息的视觉SLAM技术得到了快速发展。本文系统综述了近年国内外视觉SLAM的研究成果。首先归纳整理了视觉及视觉/惯性SLAM中的前端里程计、后端优化、回环检测和地图构建关键环节。在纯视觉SLAM技术分析方面,围绕传统基于几何变换和现代基于深度学习的两大类视觉SLAM方法开展讨论,并梳理了相关具有代表性的视觉SLAM技术。在详细阐述视觉/惯性组合导航的多信息融合SLAM方法基础上,分析了近年基于异源图像的视觉及其与惯性组合的SLAM导航技术进展。最后对视觉SLAM的未来发展方向进行了针对性展望。