Research On Positioning Technology Indoor Of UAV Based On Optical Flow

With the continuous expansion of the application scenarios of the quadrotor UAV,how to accurately locate without GPS signal in the unreliable situation is the future development trend of the UAV technology.Positioning the environment perception through the visual sensor has become the preferred solution for the precise positioning of the UAV in the indoor environment due to its low system cost and small size.Optical flow is a method for estimating the displacement of pixels in an image sequence with time.The algorithm can effectively capture the displacement deviation of the UAV,the environment without the GPS signal,such as indoors,caves,etc.It is important for UAV to fly stably.The focus of this paper is on the research of indoor UAV positioning algorithm based on optical flow.The main contents are as follows:First,the basic knowledge of optical flow is briefly introduced,including the constraints of establishment of optical flow model and the limitations of optical flow method.On this basis,the basic conditions,solutions and limitations of the classical models of sparse optical flow and dense optical flow are further elaborated.In addition,the matching criteria and search methods used in the process of building block matching optical flow models used in traditional UAV optical flow modules are also studied.Secondly,there are some problems of small amplitude pendulum motion and low robustness of feature points in the visual positioning process of indoor UAV.These problems will cause optical flow estimation to be inaccurate.In this paper,an improved LK optical flow algorithm based on improved forward and backward error compensation pyramid is proposed,which is used to track feature points.Furthermore,based on the idea of bidirectional reversible constraint,a new objective function is proposed to update the optical flow estimation offset,and the performance of the algorithm is verified by static experiments.Experiments show that the proposed algorithm effectively improves the accuracy of optical flow estimation.Then,the Pixhawk UAV is used to verify the actual flight of the proposed algorithm and compare it with the traditional block matching optical flow.In order to improve the flight stability of UAV,the current problems of optical flow algorithm are studied.The robust optical flow model which can deal with noise,large displacement and illumination changes is explored.This paper introduces adaptive median filtering based on the improved algorithm.The illumination proportional compensation method and the Hessian matrix improve the robustness of the algorithm.By usingfixed-point and pendulum flight experiments,the experimental results show that the optimized algorithm is improved in accuracy and speed.Finally,the four-rotor UAV experimental platform for indoor flight was designed and built independently.The platform was used for indoor flight experiments.During the experiment,the Optitrack motion capture camera was used to collect the real-time position and attitude data of the UAV flight,in natural light,The hover test was carried out in three scenarios of light and low light,which further verified the feasibility and robustness of the algorithm.