| At present,indoor positioning technology based on multiple sensors has been widely used due to its high precision,high reliability,great portability,and low cost.Compared with the positioning system based on single sensor,a multi-source fusion positioning system based on a variety of heterogeneous sensors can gather various positioning information for fusion,and seamlessly and accurately obtain the position of the target.In indoor scenarios,multi-source fusion positioning technology has the advantages of wide positioning coverage,high reliability,and strong anti-interference ability.This paper selects the robust multi-source fusion positioning based on multiple heterogeneous sensors and related technologies as the research goal,aiming to achieve robust multi-source localization in complex indoor environments through the optimization and fusion of data observations from multiple heterogeneous sensors.The important works involved in the thesis are as follows:1.According to the requirements for fast calibration of the positioning beacon in the indoor multi-source fusion positioning,the Simultaneous Calibration and Localization(SCAL)technology is studied.Aiming at the limited accuracy of the existing SCAL system when acquiring real-time positions of beacons and targets,a novel SCAL algorithm framework based on Error-weighted Geometric Dilution Precision(EGDOP)Optimization is designed.It consists of two parts:1)on the Target Localization and Beacon Calibration section,for the typical beacon distribution,a processing mechanism for synchronous target positioning and beacon calibration is designed to initially synchronously locate the target and the beacons;2)on the Global Optimization section,an optimization model based on EGDOP is proposed,using the geometric distribution of the target trajectory and positioning error to screen out the set of trajectory points for real-time global optimization.The SCAL algorithm framework can simultaneously acquire the positions of moving targets and beacons with high precision while the system is running in real time.2.With the Filter as the basic fusion framework,this paper studies the fusion positioning algorithm that can eliminate the influence of abnormal observation data.Aiming at the problem that the existing Huber-based fusion algorithm lacks an adaptive adjustment mechanism to the dynamic noise,which leads to the loss of fusion positioning performance,this paper proposes an Adaptive High-order Correction(AHC)cost function.The ABC function can generate a correction factor that can dynamically adjust the weight of the filtered observation according to the dynamic noise,and then adaptively suppress the abnormal observation data.On this basis,combined with the Cubature Kalman Filter(CKF),this paper proposes a multi-source fusion algorithm based on the AHC-CKF,to perform robust fusion of the positioning observation data from 3 or more kinds of sensors.3.This paper studies the "plug and play" fusion system based on the fusion filtering algorithm,which ultilizes IMU-based forecast updates and asynchronous observation updates based on multiple heterogeneous sensors(such as UWB,visual cameras,barometers,etc.).On this basis,a simulation and experiment platform for multi-source fusion positioning based on actual scenes is designed.Through multiple simulations and experiments,it is analyzed and verified that 1)compared with the existing SCAL algorithm,the proposed SCAL algorithm has higher computational efficiency,and with the premise of real-time operation,its positioning accuracy of the target and the calibration accuracy of the beacons are increased by 39.06%and 47.97%respectively,reaching 0.2557m and 0.2437m;2)compared with the existing fusion algorithm,the positioning convergence of the proposed AHC-based fusion algorithm has increased by 5.42%,and its target-positioning accuracy has increased by 19.53%to 0.1801m. |
PDF Full Text Request