Research On High Precision Positioning Of Indoor Target Based On Audio And Time Difference Of Arrival

With the development and improvement of navigation technology,location-based services are becoming more and more important in people's travel.However,the easy-to-use indoor positioning technology is not yet mature.Indoor positioning has broad application prospects in many fields,such as the positioning services for vehicles in underground parking lots,navigation services provided in large museums,accurate marketing in superstores,health monitoring for patients and elders in hospitals and nursing homes,and so on.In recent years,based on smart terminals,researchers have proposed various indoor positioning solutions based on computer vision,infrared-based,Bluetooth-based,Wi-Fi-based,and audio-based technologies.Among them,audio-based indoor positioning has attracted widespread attention due to its advantages such as low cost,high accuracy,and easy deployment.This dissertation aims to realize a high availability,high precision and robust indoor audio localization technology based on time difference of arrival(TDoA).Focusing on the basic positioning algorithm based on TDoA.Two new shrinking circle(SC)methods based on binary dichotomy are proposed.According to the positioning error caused by height and moving speed in practice,two error elimination schemes are proposed.Simulations and field experiments were conducted to compare the performance of various localization algorithms,and to verify the effectiveness of the error elimination schemes.The research results of this thesis will contribute to the application and promotion of indoor audio location system.The specific research work and results of this article can be summarized as follows:(1)Research on localization algorithm based on TDoA.Based on the conventional shrinking circle(CSC)method,two improved shrinking circle method(SC-1 and SC-2)based on the dichotomous strategy are proposed and simulation and audio-based real-world positioning experiments are performed.Performance of the traditional shrinking method,two improved diminutive strategies based on the bisection strategy,and various localization algorithms based on the least squares method are compared and analyzed.The results show that the separated constrained weighted least-squares(SCWLS)method performs best in terms of precision and robustness.Performance of the three shrinking circle methods(SC-1,SC-2 and CSC)are slightly worse than SCWLS,but better than two-step weighted least-squares(2WLS)method and constrained weighted least-squares(CWLS)method.In terms of time consumption,SC-2 shows the fastest operating speed.(2)Error analysis based on the actual positioning system.The purpose of this study was to eliminate positioning errors due to height and movement speed.Aiming at the problem of positioning error in the two-dimensional plane caused by the height difference between the plane where the speaker node is located from the target terminal,a compensation scheme using the radius projected onto the two-dimensional plane instead of the three-dimensional radius is proposed.For the problem of positioning error caused by the movement of the terminal,in this paper,the displacement of the terminal occurring during the positioning process is equivalent to the displacement occurring at the node,and the terminal is positioned using the corrected node coordinates.The effectiveness of the two error cancellation schemes are verified by simulation experiments.(3)Filtering research based on audio localization.The purpose of this study is to estimate the true position of the target from the noisy positioning results.In this paper,the performance of the extended Kalman filter(EKF),strong tracking filter(STF)and unscented Kalman filter(UKF)are studied.In the study,a linear motion model and a measurement model were constructed based on the characteristics of indoor positioning.The noise covariance matrix was constructed based on statistical characteristics of static positioning results.The results of simulation and field experiments showed that EKF and UKF performed in a consistent manner.STF showed the ability to track system state mutations.However,STF showed poor performance when the state of the system continuously mutates.