Research On Indoor Positioning And Mapping Of UAV Based On Monocular Vision And Inertia

Visual Simultaneous Localization and Mapping(VSLAM)is a key technology for robot to reconstruct environment scene based on visual sensor data in unknown environment.Although the VSLAM methods based on monocular vision sensor data have the advantages of low price and being easy to use,it has some problems,such as scale drift,large amount of calculation,fast motion and crash.In this paper,the closed-loop detection problem based on photometric direct method and the monocular VSLAM method based on inertial sensor data are studied.The problems of scale uncertainty are solved,and the experimental verification is carried out on the simulation system and public data sets.The main research work of this paper is as follows:(1)The closed-loop detection problem based on photometric direct slam method is studied.Firstly,the optimization problem is combined with Closed-loop Detection in the framework of VO algorithm based on photometric optimization,and the pose estimation results are constrained in the visibility window of local map by setting common view frame;in addition,the persistent map is constructed by creating common view map and merging local map function to deal with the closed-loop detection problem of the system and improve the accuracy.The effectiveness and feasibility of the proposed algorithm are verified by experiments on public data sets.(2)The initialization and back-end optimization of the tightly coupled visual inertial positioning system which integrates monocular image data and inertial measurement unit(IMU)data are studied.Firstly,the system is initialized,and the loose coupling method is used to calculate the vision and IMU data respectively.According to the calibrated rotation matrix,the initial value of VIO system is calculated.Then,based on the initial value,the joint optimization equation of photometric error and IMU Pre-Integration error is constructed by using the tight coupling method.Finally,aiming at the problem of the increase of system computation and the management of map points,the strategy of edge optimization is adopted to ensure the real-time and accurate operation of the system,and the feasibility and robustness of the method are verified on the open data sets.(3)The positioning system in this paper is verified by the software experimental platform and the hand-held module device integrated with camera and IMU.Firstly,the calibration experiment of the hand-held device is carried out to ensure the accuracy.Then,several groups of experiments are carried out in the indoor hand-held experimental scene and UAV flight experimental environment based on simulation software to verify the system.