The Research On Vehicle Positioning Based On Visible Light Communication

Recently,visible light communication(VLC),as a complement to Radio Frequency(RF),is becoming a relatively mature communication technology.In the filed of VLC,as vehicle VLC has become a research hotspot,the vehicle positioning method based on VLC system framework,with its advantages of green environmental protection,low cost,simple framework and high positioning accuracy,has attracted extensive attention of scholars.Among them,VLC based vehicle positioning will face different goals and challenges in different scenarios.In this paper,the VLC based vehicle positioning methods are investigated in different scenarios of vehicle communication based on VLC,and the main works are as follow:1.A vehicle to vehicle(V2V)ranging method based on VLC is proposed.The monocular ranging method is applied to outdoor vehicle positioning,and the distance between the taillight LEDs of the target vehicle is taken as a reference to effectively alleviate the scale drift of monocular ranging.And an image processing scheme is proposed to eliminate the existing image background noise and extract the vehicle taillight LEDs.Moreover,the Kalman filter(KF)is used in the ranging process to decrease the random errors,smooth the error curve and improve the positioning accuracy.In the end,the experimental results show that under the experimental conditions,the proposed method can achieve centimeter level ranging performance within the range of 15 m ～ 52.5m,and has a certain robustness to the speed change.2.A infrastructure to vehicle(I2V)positioning method based on VLC is proposed.Convolutional Neural Network(CNN)based decoding method is proposed to overcome the jello effect,which always happen when receiving signals with a moving state in the I2 V scenario;and aiming at the problem of the effective LED light source selection,the effective LED light source selection method is proposed to solve the problem;Meanwhile,in order to provide high precision positioning results,after the position information sent by the transmitter is decoded,a geometric positioning algorithm is used to obtained a more accurate vehicle world coordinate.The experimental results show that under the experimental conditions,the proposed CNN model can effectively decode the information send by LED light source,and can achieve nearly 100%decoding accuracy;Also,the proposed geometric positioning method can control the average positioning error within 6cm in the positioning range of 7m ～ 15 m.