Research On Autonomous Tracking Of Boat-borne UAV Based On Vision

Through the coordination of aerial drones,surface drones and underwater submersibles,all-round information perception can be achieved,and tasks such as maritime monitoring,environmental detection and maritime target tracking can be more effectively completed.Unmanned boats have the advantages of long range,long working hours and large loads,but their information perception capabilities are limited.The drone has a large aerial viewing angle and strong environmental awareness,but its endurance is limited.Combine the advantages of the two,overcome their weaknesses,and give full play to their respective strengths.Before the UAV’s energy is exhausted,the UAV can accurately land on the specific position of the UAV and automatically replenish the energy.It is necessary to conduct research on the autonomous tracking and autonomous landing of the UAV based on machine vision to ensure no Safe and effective cooperation between USV and UAV.This thesis studies the target tracking algorithm and the visual positioning algorithm of sea-air cooperation.According to the task requirements of the sea-air cooperation task,the tracking of specific targets at sea and the autonomous return system after the UAV completes the task are studied.Aiming at the particularity of the marine environment and the real-time tracking of targets by UAVs,an improved KCF algorithm combined with TLD algorithm is proposed as a tracking algorithm for visual positioning.A positioning method based on sea-air coordination is proposed and the improved target tracking algorithm is integrated to complete the precise positioning of the UAV on the landing platform,and the hardware platform is built for experimental verification:(1).According to the tracking characteristics of UAVs on sea targets,the advantages and disadvantages of KCF algorithm and TLD algorithm are analyzed.And according to the realtime requirements of UAV tracking,and the complexity of the maritime environment.Firstly,the KCF algorithm is multi-feature fusion and multi-scale transformation strategy is set.The improved KCF algorithm is used as the tracking module of the TLD algorithm,and the update strategy of the filtering template is determined according to the confidence of the tracking result.In order to improve the real-time performance of the algorithm,a Kalman filter algorithm is introduced to localize the detection module.Through the tracking experiments and quantitative analysis of actual sea targets,it is verified that the improved algorithm can meet the requirements of sea target tracking.The improved tracking algorithm is applied to the DJI M100 UAV.In the tracking of small-scale moving targets,the position of the target is obtained through the visual tracking algorithm.The center position of the field of view;for a large-scale moving target,on the basis of solving the relative position through the visual tracking algorithm,the UAV tracking controller is designed by using the PID control algorithm to control the UAV to track the dynamic motion of the sea target,and monitor the PTZ.Tracking and dynamic tracking were performed with actual experiments.(2).Aiming at the problems of large positioning dead zone and complex marine environment in the process of autonomous landing,according to the current airborne vision landing technology,a positioning method based on sea and air dual-camera visual fusion navigation is proposed,supplemented by an improved target tracking algorithm.The positioning algorithm,and the Ar Uco code with error correction function is designed as the positioning identification code.Combined with the target tracking algorithm to select the ROI area,detect the identification code to obtain the positioning information,which can effectively improve the real-time recognition and solve the instability of the identification code identification.The target positioning algorithm combined with Kalman filtering is used for position estimation and correction to realize the discrete data.Continuous,to prevent instability caused by sudden changes in data,and to smoothly track the landing process.On the basis of visual positioning,combined with PID control algorithm,the UAV autonomous landing controller is designed,and the DJI M100 UAV is selected as the carrier to conduct practical experiments on the landing control system.The effectiveness of the improved algorithm for target tracking at sea,the accuracy of dual vision co-location,and the feasibility of the tracking system and autonomous landing system are verified through experiments.