Based On Computer Vision-assisted Navigation System

Comparing to the general airplanes, the UAV has the advantages of low cost, convenience of transportation and maintenances, furthermore, there is no need to worry about the casualty problems, Therefore, the UAV can be widely used in military and civil field. But the recovery of the UAV can't be satisfied, thus it limits the application to all fields.This thesis takes the GPS/SINS integrated scheme which is hot domestically and abroad for reference, and then brings forward the scheme of utilizing the way of Kalman filter to combine the information from computer vision and SINS to navigate the landing procedure of the gauzy fixed unmanned plane.During the period of designing the Kalman filter, we adopt the feedback correction mode. The input of Kalman filter is the differential of the two position values as follows: one is the position information of ground surface feature point that is captured in the image by the computer vision system;the other is the position information of the ground surface feature point in the projection surface that is calculated with the position and posture information of the UAV from SINS by the video camera model. After filtering, the output of Kalman filter is the estimation of inertial navigation deviation, and then we can calibrate the SINS.After simulating the entering procedure, we find that computer vision can capture a higher resolution image, even though the precision of the gyroscope and accelerometer is low, it can still get good locating precision. In the process of simulation, by adjusting the precision of the gyroscope, accelerometer and video camera, we can get a series of results. The comparison of the results shows that the resolution of video camera affects the locating precision most, the gyroscope does less, and the accelerometer does the least. Furthermore, the influence of video camera resolution to the locating precision expresses exponent relation;when the precision reaches a certain scope, continuing increasing the resolution of video camera can do little to the precision.The result of the simulation shows that the scheme from this thesis can increase the entering precision of unmanned plane in certain degree, and there is some practical value.