Research On Visible Light Indoor Positioning Technology Based On Multi-Source Information Fusion

With the rapid development of 5G and the popularity of mobile intelligent terminals,more and more personalized applications are emerging,among which the demand for navigation and location services is increasingly urgent.Indoor positioning technology provides an extremely important guarantee for indoor navigation and location services.an indoor positioning technology.As an indoor positioning technology based on visible light communications(VLC),Visible light positioning(VLP)uses the ubiquitous light-emitting diode(LED)as an active beacon for positioning.It can provide a new reference value for the realization of low-cost and high-precision indoor positioning.The topic of this thesis comes from the project of "Research on light and visual information combined high accuracy visible light localization theorem and technologies" supported by National Natural Science Foundation of China(Project Number:61901047).To solve the problems of limited transceiver orientation and insufficient coverage faced in VLP technology,this thesis innovatively combines traditional single sensor information and proposes two VLP algorithms that fuse different sensor information,which significantly improves the availability and coverage performance of VLP algorithms.The main research contents of this thesis are following:(1)Aiming at the problem of insufficient availability due to the limitation of both LED orientation and receiver orientation in VLP,a camera assisted received signal strength(CA-RSS)algorithm for VLP is proposed by effectively combining the visual and strength information of light source,which not only does not need to limit LED orientation and receiver orientation,but also can achieve high-precision positioning.Based on rectangular LED layout,CA-RSS algorithm utilizes the linear least square method and solid geometry theory to achieve two-dimensional(2D)and three-dimensional(3D)positioning,respectively,which effectively solves the error propagation issue.Meanwhile,CA-RSS is not limited by the orientation and the height of LEDs when estimating the irradiance angles,and thus it can be applied to more diversified indoor scenarios which greatly improves the availability.Simulation results show that CA-RSS can achieve 90th percentile accuracies of about 12.5 cm for 3D positioning regardless of the receiver orientation,LED orientation and LED height.In the process of experimental verification,considering the additional error caused by the imperfection of hardware and channel model in the actual environment,a model calibration method is further proposed.The actual channel model is fitted using the measured received signal strength(RSS),and then the parameters are calibrated compared with the theoretical model.Experimental results show that using the model calibration method,CA-RSS can achieve an average 3D positioning accuracy of 6.31 cm with different receiver orientations.(2)Aiming at the insufficient coverage performance caused by the limited number of LEDs in VLP,an attitude sensor assisted single view geometry(ASA-SAG)algorithm for VLP is proposed by effectively combining the visual information of light source and the attitude information of receiver,which can achieve high-coverage and high-accuracy positioning.Based on single view geometry theory,coordinate transformation and plane geometry theory,ASA-SAG establishes the conversion relationship of LED and LED’ projection in different coordinate systems,and thus can achieve single LED for 2D positioning and double LEDs for 3D positioning,which not only ensures the positioning accuracy of the algorithm,but also greatly improves the coverage of the algorithm.Simulation results show that ASA-SAG can achieve more than 90th percentile coverage.In addition,experimental results show that ASA-SAG can achieve an average 3D positioning accuracy of 5.54 cm with different receiver attitudes,and can also achieve stable positioning in a mobile environment.