Monocular Vision SLAM Location Method Of Underground Unmanned Aerial Vehicle For Industrial Internet

Through the underground industrial Internet,underground Unmanned Aerial Vehicle(UAV)and other intelligent devices can be remotely managed which great significance to achieve unmanned mining.In order to manage underground UAV,the technology of UAV autonomous positioning and navigation in mine environment needs to be realized.Simultaneous localization and mapping(SLAM)algorithm can position UAV accurately by using sensors to monitor around environment.Therefore,the underground mine UAV vision SLAM algorithm is studied to realize its autonomous positioning and navigation.The main contributions of this thesis are shown below:(1)For wide obstacle-free roadway,a two-dimensional code with location information is set up on the top wall of the roadway as the guiding sign of UAV.Then for narrow roadway,a reflective identification plate is set up on two sides of the roadway to be the guide sign.For these two different roadway environments,geometric-topology based underground roadways are created respectively.(2)Aiming at the wide roadway environment,a two-dimensional code with position information is proposed.According to the edge detection and the straight line fitting algorithm,the code information of this two-dimensional code is obtained.Simulation results show that this algorithm can identify the two-dimensional code fast and clearly.Aiming at artificial signs in the narrow roadway environment,the reflective identification plates and the surrounding natural features are acquired and stored in library.An SIFT algorithm based on RANSAC is proposed to extract the features of each image captured by UAV,and to match with the pre-established feature image database.Simulation shows that the SIFT algorithm based on RANSAC has a high correct matching rate,and the landmark location information can be obtained according to the feature image database and off-line map.(3)For the wide obstacle-free roadway and two-dimensional code UAV guided landmark scene,a mine UAV monocular vision PSOFastSLAM algorithm based on two-dimensional code is proposed.Simulation results show that this algorithm improves the particle degeneration problem of FastSLAM localization algorithm and improves the positioning accuracy of UAV.(4)For the narrow roadway and reflective identification plates UAV guided landmark scene,an underground UAV monocular vision EKF-SLAM algorithm based on reflective identification plates is proposed.The position of UAV is estimated by the observing information from known landmarks.Simulation results show that this algorithm can locate the UAV accurately.The final results show that according to different roadway environments,different SLAM algorithms can be used for precise positioning of UAVs for industrial Internet.This study laid foundation for effective management for the follow-up industrial Internet to collect environmental data by using UAV.