Research On Visual Landing Control Of UAV Based On Dynamic Target Tracking

In recent years,autonomous landing of unmanned aerial vehicles has gradually become a research hotspot.The autonomous landing of the UAV can not only improve the autonomous performance of the autonomous flight of the UAV,but also improve the intelligence level of the UAV.Autonomous landing of UAV is the much complex task in all kinds of autonomous flight tasks.It not only requires UAV to earry out stable attitude control,but also requires UAV to complete high-precision positioning,which has become an important research content in the field of autonomous landing control of UAV.Aiming at the inaccuracy of traditional positioning accuracy for dynamic targets such as GPS and inertial navigation,an autonomous landing system of UAV based on dynamic target tracking by using visual navigation to realize the autonomous precise positioning of UAV is designed in this paper.The main contents of this paper are as follows:Be aimed at accurate identification of tags,based on the recognition algorithm of ArUco codes,the recognition of markers is realized through five processes:image graying,image adaptive thresholding,contour extraction,quadrangle screening,perspective change and internal id recognition.The unstable situation of marker recognition caused by illumination outside of ArUco codes is solved,and the maximum recognition height of markers and the recognition rate of markers at different heights are improved.Be aimed at targeting issues for dynamic tags,PnP visual location algorithm is applied to the positioning problem of quadrilateral markers.The problem of vision-based dynamic positioning of UAV is successfully solved.The relative position of UAV relative to the landing platform is obtained,which provides more precise location information for the autonomous landing of UAV.Be aimed at tracking issues for dynamic tags,the traditional Kalman filter is improved.An improved Kalman filter is designed to accurately estimate the motion trajectory of the target,which improves the prediction accuracy of the Kalman filter and solves the problem of the UAV's flight disorder caused by the short-term target loss in the process of autonomous tracking and landing.Finally,based on visual navigation algorithm and incremental PID algorithm,the UAV position tracking controller and visual landing controller are designed to build an experimental platform for UAV autonomous landing system based on dynamic target tracking,and flight test experiments are carried out.The effectiveness and feasibility of the UAV visual landing control method based on dynamic target tracking proposed in this paper is presented.The starting height and tracking accuracy of the UAV's autonomous landing are improved,which provides a kind of UAV's autonomous landing programme.