摘要
针对无人机GPS/SINS组合导航系统中航向角误差可观测性差的问题,提出了一种无人机GPS/SINS/Vision组合导航方法。使用无人机搭载的摄像机获取周围景物的图像序列,通过跟踪图像序列中提取的特征点,利用极几何约束获得其间摄像机的位移信息,将测量到的位移信息融合到GPS/SINS组合导航系统中,构成GPS/SINS/Vision组合导航系统。通过观测位置误差、速度误差和摄像机位移在像平面上的投影误差,建立组合导航误差状态卡尔曼滤波器,估计导航系统的误差状态,并根据估计结果对导航系统状态进行修正。推导了滤波器状态方程和测量方程,使用Simulink对算法进行仿真验证,在水平姿态角偏差为10°左右、航向角偏差为20°左右的初始条件下,水平姿态角误差和航向角误差均能快速收敛,水平姿态角误差在10s后收敛到1°左右,航向角误差20 s后收敛到2°以内。仿真实验表明,该方法能有效提高无人机姿态角估计的速度和精度。
To solve the problem that observability of yaw error is poor in GPS/SINS integrated navigation system of UAVs（Unmanned Air Vehicles）,a GPS/SINS/Vision integrated navigation system is proposed.The camera on board takes a sequence of images.Several feature points are extracted from these images.The direction of camera translation in that period is obtained by tracking these feature points.A GPS/SINS/Vision integrated navigation system is established by including the information of camera translation to GPS/SINS fusion filter.The state error of navigation system is estimated by measuring position error,velocity error and the projection error of camera translation.This approach is realized via a Kalman filter.The UAVs motion state is corrected by using estimation results of Kalman filter.In the end,a simulative experiment is made by using Simulink.Assuming that the initial horizontal attitude error of UAVs is about 10°,and the yaw error is 20° around,the horizontal attitude error declines to 1° in 10 s,and 2° in 20 s for yaw error.The simulative experiment results reveal the proposed method can effectively increase the estimation speed and accuracy.
出处
《中国惯性技术学报》
EI
CSCD
北大核心
2011年第4期457-461,共5页
Journal of Chinese Inertial Technology
基金
"十一五"国防预研基金项目(51325050203)
