| With the gradual emergence of advantages such as high load and long endurance,industrial drones have been widely used in river inspection.As a supporting infrastructure for drones,drone airports provide functions such as recycling,storage,and automatic charging for drones.Some industrial drones have wingspan lengths of over 2m and have large volumes and weights.How to safely recover them is a problem in engineering applications.This article takes the application of industrial drones in the Yellow River inspection system as the background,and proposes a scheme for the safe recycling of drones.The drones randomly land within the allowable range,and then are lifted and recycled by an Automated Guided Vehicle(AGV).Based on this scheme,this article studies the AGV navigation method.At present,research on AGV navigation mainly focuses on indoor scenes,and indoor navigation methods are relatively mature.However,due to the complexity of outdoor environments,autonomous navigation of outdoor AGVs is more difficult.In response to the complex outdoor working environment of AGV,this paper designs an AGV outdoor navigation system based on a mixed use of multiple navigation methods,and focuses on researching the key navigation methods in the system.The main contents of this paper are as follows:(1)By analyzing the AGV workflow in the Yellow River channel inspection system,the AGV navigation process is divided into two stages:coarse positioning and fine positioning:the AGV search for drones is the coarse positioning stage,and the AGV search at the bottom of the drone is the fine positioning stage.In view of the complex outdoor environment,an AGV navigation scheme which can switch the navigation mode is presented,and the main technology of the scheme is determined.(2)In the coarse positioning stage,the multi-sensor data fusion positioning system in complex outdoor scenes is studied to overcome the limitations and vulnerability of singlesource sensor positioning in outdoor environment.Using Extended Kalman Filter(EKF)as the basic fusion algorithm,the best estimation of AGV position is obtained by weighting the gain factor using the measured values from the AGV kinematics model and Real-Time Dynamic(RTK)carrier phase differential positioning based on wheel odometer.To solve the timevarying problem of observation noise in the iteration process,an EKF algorithm based on fuzzy control(Fuzzy Control,FC)is presented,which reduces the impact of observation noise changes on the estimation results in outdoor environment.The positioning and orientation accuracy are improved by 66.7%and 50%,respectively,by experimental simulation.(3)In the stage of fine positioning,aiming at the problem that the drone occludes RTK antenna,which leads to lower positioning accuracy,the single source visual navigation system based on AprilTag is studied to replace the integrated navigation system.To solve the problem that traditional AprilTag algorithm is prone to misrecognition,this paper presents an improved AprilTag algorithm based on Random Sample Consensus(RANSAC)and Least Squares(LS),which improves the accuracy and robustness of visual positioning.The experimental results show that the improved algorithm can remove 87.5%of the errors.(4)Designing the hardware and software framework of AGV control system and building the AGV entity structure according to the practical application requirements and communication mechanism.The AGV outdoor navigation system is tested in the actual application environment,and the test results show that the success rate of the system can reach 99%.At present,the research results of this paper have been successfully applied to actual projects. |
