Real-Time High Precision Localization Based On Visible Light And Multi-Sensor Fusion

With the development of the mobile internet era,people's demand for seamless positioning with higher accuracy is increasing day by day.The Location Based Service required by people is not limited to outdoor but indoor,so indoor positioning technologies have also been developed rapidly.Traditional indoor positioning technologies based on Wi Fi,Bluetooth and Geomagnetic signal have their own disadvantages,such as large indoor positioning error and interfered by external factors easily.Indoor positioning technology based on Ultra-Wide Band(UWB)has the advantage of high positioning accuracy,but the hardware equipment investment is considerable and the deployment is difficult.With the development and maturity of LED(Light Emitting Diode)technology,LBS(Location Based Service)can be realized at the same time as lighting.LED has the advantage of signal stability and not easy to be interfered during transmission.Therefore,indoor positioning technology based on visible light will be one of the development trends of indoor positioning in the future.The construction of this system is simple relatively.In this paper,the range positioning based on the trained propagation model and sensor fusion are studied in depth.The work concerned are listed as follows:(1)The construction of visible light positioning system based on external frequency regulation: using microcontroller to drive multiple LED light sources and the stable MOSFET switching circuit for LED lights to meet specific frequency flicker firstly,signals can be received by the light sensors and smart phone with 3.5mm audio port,then the Andorid smartphone estimate distance through FFT algorithm and lambert model.At last,the least squares algorithm is used to estimate the location.After the hardware positioning platform is built completely,the optical signal is received through the signal receiver.A real-time positioning display application is implemented on the mobile phone.The average positioning accuracy can be less than 0.3m under the illumination condition.(2)Position estimation based on multi-sensor fusion: through further research of indoor positioning technology based on visible light,it is concluded that the coverage of visible light is limited.For the optimal position estimation,this paper proposes a method of using Kalman filter to fuse visible light data with Pedestrian Dead Reckoning(PDR)trajectory data,which is combined the step counting algorithm based on peak detection algorithm of acceleration and the heading angle estimation algorithm based on the gradient decent algorithm of three-axis acceleration,gyroscope and magnetometer signal.That can realize seamless real-time positioning,the visible light signal can correct the cumulative error of PDR effectively.Besides,the performance of this real-time positioning APP is excellent,so the location can be calculated and displayed in real time.Through the verification of the actual indoor positioning experiment scene,the accuracy of the real-time positioning system based on visible light below 0.3m.The estimated positioning accuracy of PDR is 1.95 m,the integrated positioning accuracy based on visible light and PDR is 0.89 m.