Research On UAV BDS-RTK Attitude Determination And Location Algorithm Aided By Optical Flow

With the progress of science and society,the measurement accuracy of UAV and outdoor unmanned equipment for their attitude information is also increasing.In the challenging environment,a single attitude measurement method often cannot meet the actual mission which needs low cost,high accuracy and high reliability,so the combination of multiple attitude detection methods which is used to improve the robustness and the accuracy of the system is the most practical means at present.Therefore,this paper studies the RTK attitude measurement and positioning algorithm which assisted by image optical flow.On the basis of in-depth study of RTK Positioning and optical flow attitude estimation,the residual-base adaptive Kalman filter is used to combine the two ways to realize the comprehensive estimation of the attitude,which effectively overcomes the problems of error accumulation in the optical flow attitude estimation and the interference of RTK attitude measurement.The main research contents and innovations are as follows(1)BDS-RTK relative positioning technology and the RTK-base attitude measurement technology are studied.The conversion between coordinate systems and attitude calculation method are deeply studied.According to RTK-base attitude technology,a RTK-base attitude measurement module based on STM32F7 and UBLOX is designed and implemented.(2)By comparing and studying a variety of optical flow algorithms,the sparse optical flow algorithm based on Speed-Up Robust Features(SURF)is selected according to the needs of the task.In order to increase the number of feature points in the shadow and texture blurred regions,the CLAHE is introduced to adjust the image contrast and highlight the texture details.This method effectively improves the performance of SURF-base optical flow algorithm,and effectively increases the accuracy of image optical flow.Compared with the traditional SURF-base optical flow method,the average endpoint error is reduced by0.172 pixels,and the average angle error is reduced by 3.907°.(3)In this paper,the method of attitude solution based on optical flow is studied,and the computational complexity of attitude solution is simplified by introducing the focus of expansion.Aiming at the problem that the error of the optical flow has a significant impact on the accuracy of the attitude estimate,based on the cost function,the idea of robust estimation is used to eliminate the data with large error in the image optical flow.This method reduces the outliers in the process of angular velocity estimation,and improves the estimation accuracy effectively.The RMS of angular velocity estimated by the algorithm is in the range 0.3°/s-0.5°/s in various road scenes.(4)In this paper,the multi-sensor data fusion algorithm is deeply studied.Considering with the characteristics and disadvantages of RTK-base attitude measurement and optical flow-base attitude estimation,a integrated attitude estimation algorithm based on Kalman filter is designed to realize the data fusion of the two ways.To improve the accuracy of the KF-base integrated attitude estimation algorithm in the challenging environment,a integrated attitude estimation algorithm based on residual-base adaptive Kalman filter is designed.The system observation error is estimated in real time through residual vector,which ensures the attitude estimation accuracy of the algorithm in the challenging environment.Under the simulation conditions,the RMS of pitch,roll and heading angle is0.2670°,0.2646° and 0.2528°.