Research On Key Technologies Of UAV’s Track Deployment And Accurate Landing

With the rapid developing technology of unmanned aerial vehicles,the application of UAVs is gradually promoted both in the industry and people’s daily life.UAV transportation will become an important branch of the future transportation system.The key technology of UAV path planning and UAV precision landing are of vital importance to the application of UAV transportation.The research of intelligent algorithms based on machine learning has made significant breakthroughs recently,It brings the opportunity to improve the state-of-the-art techonologie in the research field of UAV.This paper study on the trajectory planning and image based precise landing.The two important contents of the urban UAV transportation process are the path planning and the precise landing.This article explores the key technologies of UAV trajectory planning and precise landing.It mainly focus on the path planning algorithm and image processing based landing guidance.The problem of UAV path planning is similar to that of traveling salesman problem.The idea of TSP is introduced to find the shortest path.The addressing process is transformed into the process of solving the weighted Hamilton graph,and the optimal track is calculated using the ant colony algorithm.Aiming at the traditional ant colony algorithm,a marker improvement method based on Newton’s cooling method is proposed.The results of simulation and experiment prove that the track length of the unimproved algorithm is 1359.8m,and the optimal track length of the improved ant colony algorithm is 1337.9m.The improved ant colony algorithm is better than the traditional ant colony algorithm,the addressing is more suitable for practical applications,and can effectively solve the problem of UAV path planning.At present,the guidance of UAV landing mainly relies on GPS and.Beidou navigation.In the absence of GPS or Beidou navigation and positioning,the reference frame needs to be re-established for position calibration.UAV visual positioning is a method to assist UAV landing.First,based on the scale-invariant feature transformation,the feature points of the image information and the actual geographic location are found,and then the location near the landing site is found based on the feature matching.Aiming at the control of the landing attitude of the UAV,it is proposed to use the edge of the image to detect the contour of the characteristic position.According to the strategy of finding the distance and azimuth of the image center,the UAV attitude adjustment parameters are calculated,and these parameters are further called by the ground station to control the UAV to move to the landing site.The experimental results show that the average Euclidean error of this method is 5.4cm,and the horizontal and vertical axis errors are 3.2cm and 3.7cm,respectively.In the absence of GPS or Beidou navigation,precise landing of drones can be achieved.This dissertation studies the key technologies of path planning and precise landing during UAV transportation,and proposes corresponding improvement methods for specific problems.Finally,through actual flight experiments,it is shown that the method proposed in this paper can be very effective in practice and produces improved results compared with the previous works.