基于因子图的无人机视觉定位方法

UAV visual positioning method based on factor graph

当前多用途无人机主要依靠全球导航卫星系统(GNSS)和惯性测量单元(IMU)的组合导航实现高精度定位,然而在强电磁干扰以及各类复杂环境GNSS拒止等情况下,该定位模式面临失效的风险。为探索一种GNSS拒止环境下的导航定位手段,对视觉导航方式进行了研究,针对无人机小交会角条件下视觉定位存在高程方向交会精度差、绝对尺度信息缺失、累积误差无法消除和定位轨迹不连续等问题,提出了基于因子图融合影像匹配和视觉信息的光束法平差方法,解决了无人机小交会角下的自主绝对定位问题。

At present,multi-purpose UAVs mainly rely on the combined navigation of Global Navigation Satellite System(GNSS)and Inertial Measurement Unit(IMU)to achieve high-precision positioning.However,in the cases of strong electromagnetic interference and GNSS rejection in various complex environments,this positioning mode faces the risk of failure.To explore a navigation and positioning method in the GNSS rejection environment,the visual navigation method is studied.To solve the problems of poor intersection accuracy in the elevation direction,lack of absolute scale information,inability to eliminate accumulated errors,and discontinuous positioning trajectory in visual positioning under the condition of small intersection angle of UAV,a bundle adjustment method based on factor graph fusion image matching and visual information is proposed.The problem of autonomous absolute positioning under the small intersection angle of UAV is solved.