Indoor Navigation System Of Quadrotor Research Based On Vision

The micro quadrotor has the characteristics of simple structure,strong maneuverability,low cost and flexible control.It is safer than ordinary quadrotor because of its small size and light weight.The micro quadrotor is of great significance and high in application value.At present,most control systems of commercial quadrotors use GPS technology to navigate,but in some complex environments,such as urban buildings,jungles or indoor environment,satellite navigation information is not reliable.This paper mainly studies the navigation technology of quadrotor in indoor environment.In this paper,the pose estimation and positioning navigation problems of micro quadrotor mainly delying on the binocular stereo vision sensor were studied.The main work and innovations of this paper are as follows.First of all,several commonly used indoor navigation methods are studied and compared in the paper.Then the binocular stereo vision technology is choosed for indoor navigation after considering all aspects of factors.The pose estimation methods of quadrotor based on visual sensors can be divided into two categories:one is pose estimation methods based on airborne visual sensors and the other pose estimation methods is based on external visual sensors.As the image of airborne vision is susceptible to factors such as jitter during flight,the image noise is large and the image quality is not very good.In contrast,the pose estimation methods based on external visual sensors are usually more reliable and more accurate.In this paper,tthe quadrotor navigation system is consisted of an external binocular stereo vision system,an airborne IMU,a barometer and other components.Secondly,the flying principle of the quadrotor is introduced,and the dynamic modelling of the quadrotor is carried out.The double-loop control structure with an attitude inner control loop and a position outer control loop is designed and the tradtional PID control method is used to design the attitude control law and the position control law of the quadrotor.The simulation experiments of the spot hover control and trajectory tracking control are carried out to verify the effectiveness of the proposed algorithm in Matlab/Simulink.Then,the theoretical system of binocular vision system is introduced.Based on the imaging principle of pinhole camera,the binocular cameras are calibrated using Zhang Zhengyou camera calibration method and Brownian distortion model,and the internal parameters of both the left and the right cameras are obtained.Then the binocular cameras are stereo calibrated,the rotation matrix and the translation vector of the right camera coordinate system relative to the left camera coordinate system are acquired.Then,the pose estimation of quadrotor,which is the basis and key to realize the quadrotor's positioning and navigation is studied.In this paper,binocular vision sensors are used to identify the visual markers on the quadrotor and reconstruct the three-dimensional position information in the world coordinate system.According to the relative position between the visual markers,an improved orthogonal iterative algorithm is used to reconstruct the attitude information of the quadrotor in the world coordinate system.The improved algorithm has reduced the running time as well as improved the accuracy to a certain extent.Furthermore,a new pose estimation method based on the fusion of visual information and airborne IMU information has been proposed in the case that one of the visual markers can not be observed.Finally,the indoor navigation experiments of the micro quadrotor have been done.Then the feasibility of quadrotor indoor navigation system based on external binocular vision is verified.