Research On Simultaneous Localization And Mapping Algorithm Based On Visible Light

With the development of mobile Internet and the popularization of smartphone,people has increasing demand on indoor Location Based Services(LBS),and indoor positioning plays a key role in LBS.Traditional indoor positioning technologies have problems of low indoor positioning accuracy and poor stability,such as indoor positioning technologies based on WiFi,Bluetooth,and geomagnetism etc.Emerging indoor positioning technology,such as Ultra-Wide Band(UWB)indoor positioning technology,has high positioning accuracy,but the equipment is expensive and the deployment is complex.As a new generation of light sources,Light Emitting Diode(LED)is becoming increasingly popular in indoor lighting,providing ready-made infrastructure for indoor positioning in visible light.As a positioning medium,visible light has the advantages of stability and strong anti-interference.Therefore,indoor positioning based on visible light has the advantages of high positioning accuracy and strong universality,which is one of the development trends of indoor positioning in the future.This paper focuses on fingerprint positioning and ranging positioning based on visible light and the main contents are as follows:(1)In order to improve the efficiency of signal fingerprint map construction and update,this paper proposes a rapid construction scheme of signal map based on Visible Light GraphSLAM(VL-GraphSLAM)algorithm.In this part,we mainly do the following work.Firstly,we design visible light fingerprints based on external frequency control and FFT algorithm.Then,we propose VL-GraphSLAM algorithm to optimize the original Pedestrian Dead Reckoning(PDR)trajectory,and propose an enhanced GraphSLAM front-end and a robust GraphSLAM back-end to improve the accuracy of trajectory optimization.Finally,we propose to anchor the estimated trajectory in floor map according to doors as our landmarks to restore the walking path.The visible light fingerprint on the walking path is marked by the anchored trajectory to build a signal fingerprint map.(2)This paper has studied the indoor positioning technology based on visible light.Firstly,We use the WKNN algorithm to implement the positioning based on the rapidly constructed visible light fingerprint map.Then in terms of the problem of limited visible light coverage,we propose a fusion-based positioning algorithm based on Kalman filter to fuse visible light fingerprinting and inertial sensors to accurately locate the users.Aiming at the problem that the visible fingerprint can only locate the position on the walking path,we implement visible light ranging based on the trained propagation model,estimate the indoor LED position using a visible light fingerprint map,and then use trilateration technology to achieve indoor full coverage ranging positioning.After experimental verification of actual indoor scenes,the VL-GraphSLAM algorithm proposed in this paper can recover the walking path with an accuracy of 0.4m.The fingerprint positioning,fusion positioning and full-range ranging positioning based on the visible light fingerprint map constructed based on the VL-GraphSLAM algorithm achieve average positioning accuracy of 0.8m,0.9m,and 0.5m,respectively.