High Precision Seamless Positioning And Navigation For Vehicles Based On Radar Measurements

In the vehicle network,the intelligent vehicle needs to have the functions of environmental perception,path planning,autopilot and so on.Among them,the high precision positioning of the vehicle is the basis and key of the upper layer service of the vehicle network.Standard global positioning system(GPS)provide an accuracy of 5-10 meters in open sky areas,if there are obstructions such as urban high-rise areas,overpasses and tunnels,the performance will be worse.Inertial measurement unit(IMU)has the characteristics of independent of external information and high accuracy in short time,but it has serious error accumulation characteristic,which makes it impossible to be used alone.Aiming at these problems,this paper proposes the implicit cooperative positioning(ICP)method of vehicles,which breaks the traditional method of vehicle positioning,and further improves the positioning accuracy through radar ranging and angle measurement and position sharing between vehicles.Taking full advantage of the advantages of high reliability,high mobility and enhanced coverage of unmanned aerial vehicle(UAV),UAV is used as an auxiliary relay node to improve the positioning accuracy of UAV,so as to realize the requirement of full coverage of vehicle positioning in vehicle networking.Because of the high mobility and uncertainty of intelligent vehicle,the nonlinear of vehicle radar measurement equation and the non-Gaussian of ranging noise,the computational complexity of traditional filtering methods increases exponential,and the standard belief propagation algorithm is no longer applicable.Therefore,a method based on non-parametric belief propagation is proposed in this paper,which can effectively overcome the computational complexity of multi-sensor fusion and non-Gaussian noise simulation.In order to avoid the concentration of computing resource s,this paper also proposes a distributed method,the method of vehicle-vehicle cooperation.At the same time,in order to explore the lower limit of the system's positioning accuracy and to judge the validity of the system's location,the Cramer-Rao Lower Bound(CRLB)of the system is derived.With the maturity of large-scale antenna array technology,the resolution of radar for object detection is also improving.Therefore,vehicle tracking is no longer limited to point source targets,but more inclined to track extended targets with a certain spatial structure.Considering the large number of vehicles on the road,this paper proposes a multi-vehicle contour tracking based on random matrix,which is mainly divided into two steps: the first step needs to be processed by density clustering algorithm,and then the vehicle observation cluster is obtained by Gaussian mixture clustering algorithm,and the second step is to obtain the correct extended target tracking through the correct data association algorithm.In this paper,based on the above two research contents,the feasibility of the proposed algorithm is derived from the system modeling,the lower limit derivation and the simulation verification,which lays the foundation for the high-precision positioning of the vehicle in the future.