The Attitude Information Extraction Method Of MAV Based On The Visual Aid Navigation

In this paper, in order to solve the problem of MAV’s attitude parameters extraction,considering the problems of common navigation methods, such as INS’s error drift andGPS’s restricted applications, there are some limitations if they are used in MAV. For thissituation, using the visual navigation as an aid navigation, in case of the failure of othernavigation methods, assist to extract MAV’s attitude information..Firstly, we study the method of extracting the feature straight line horizon from airborneimages. Perform edge detection on the dark channel image, which is the by-productobtained during the single image haze process, and then combine Hough line detection,least squares method and rotation method to get the horizon line equation; On the basis ofobtaining the horizon line parameters, use the projection imaging relationship of the linearcamera, to deduce the relationship between the linear equation of horizon in the imagecoordinates and the MAV’s attitudes, and then calculate the attitudes using the horizonparameters: the roll angle and the pitch angle, and verify the correctness of the algorithmthrough the experiment.Secondly, after taking the algorithm simulation verification in the Matlab environment,the algorithm is ported to DM642for the real-time hardware implementation. Based on therequirements of the visual processing system, DM642hardware visual processing system isbuilt; and underlying the hardware platform, the software of the visual processing systemis designed. The results show that the algorithm is running feasibly and in real time.Then, MAV’s visual navigation semi-physical simulation system is designed, simulateMAV’s flight dynamic scenes in the air under the laboratory environment, to verify theperformance of visual processing system and the attitude parameters extraction algorithm.Finally, we carry out the visual navigation semi-physical simulation experiment underthe laboratory environment, simulate MAV’s flight path and attitudes change, then extractthe attitudes using the attitude extraction algorithm. Experimental results show that the useof the visual navigation semi-physical simulation system as well as the attitude parametersextraction algorithm can accurately obtain the linear parameters of the horizon, and then getthe attitudes with high accuracy angle, and has good real-time.