摘要
针对微小型无人机(UAV)自主运动状态估计问题,提出了一种基于双目视觉和惯性器件的微小型无人机运动状态估计方法,运动状态估计系统由微惯性测量单元(MIMU)和2个摄像机组成。根据微惯性测量单元的输出更新无人机运动状态,同时由摄像机获得周围环境的图像序列,从序列图像中提取特征点,并对这些特征点进行匹配、跟踪,测量序列图像中特征点的成像位置,建立卡尔曼滤波器,通过观测特征点成像位置误差估计无人机运动状态误差和特征点距离误差,并由误差估计结果修正无人机运动状态和特征点位置。最后,通过仿真表明,在飞行高度为50m左右的情况下,使用分辨率为800像素×600像素、基线为0.6m的双目视觉系统,实现了特征点距离估计和无人机运动状态修正,对距离为150m左右的特征点,距离估计相对误差为3%左右,速度估计误差为1m/s左右,实现了在较大场景中,使用双目视觉和微惯性测量单元的微小型无人机自主运动状态估计。
To estimate the movement of micro unmanned aerial vehicle（UAV） independently and accurately,a new method for estimating the micro UAV states of motion is presented in this paper using a micro inertial measurement unit（MIMU） combined with a stereo camera pair.The movement of a micro UAV is deduced using the measurements of the MIMU while two sequences of images are taken by the stereo camera pair,from which several feature points are extracted and matched.The movement of the micro UAV and the position of these feature points are estimated by measuring the position of these feature points in sequence images.This approach is realized via a Kalman filer.Finally,a simulation is performed by assuming a micro UAV flying at a height of 50 m with a stereo camera pair whose image resolution is 800 pixel×600 pixel and whose baseline is 0.6 m.For the feature points about 150 m away,the error percentage of distance estimation is around 3% using the method proposed in this paper,and the error of velocity estimation is about 1 m/s.The results of the simulation demonstrate that the method in this paper is efficient even in a comparatively large field of environment.
出处
《航空学报》
EI
CAS
CSCD
北大核心
2011年第12期2310-2317,共8页
Acta Aeronautica et Astronautica Sinica
关键词
微小型无人机
双目视觉
特征点
微惯性测量单元
运动状态估计
micro unmanned aerial vehicle
stereo vision
feature point
micro inertial measurement unit
motion state estimation
