Research On Multi-sensor Fusion Positioning Method Based On Extended Kalman Filter

The problem of intelligent vehicle positioning is the first problem in the field of driverless cars.It is the first step to realize autonomous navigation of intelligent vehicles and the most important step to realize intelligent transportation.At present,global satellite positioning system is the most mature technology of intelligent vehicle positioning method,which can provide users accurate location information all-weather,omni-directional,and anytime.,but in a relatively closed place or area blocked by other objects,positioning system noise,GPS receiving signal is not stable,lead to locate or distortion is not allowed.Inertial navigation is also a kind of commonly used in positioning method,it can provide high precision positioning information does not rely on outside information,noise is small,but because of time there is a drift,positioning information through the integral of time will produce the error.In view of the above deficiencies of GPS and INS,this paper proposes an extended kalman filter(EKF)GPS and INS information fusion location algorithm.Firstly,we establish the INS system model and GPS observation model for intelligent vehicles.Based on the initial position and error of GPS,we estimate the position and velocity of INS,and then derive EKF fusion algorithm.Considering that the observation value of the fusion system is affected by the noise in the actual driving process of the intelligent vehicle,we propose the EKF fusion algorithm under different state constraints of GPS and INS in order to further optimize the observation value.Through the ideal observation method,we introduce linear and nonlinear equal-constraints so that the fusion positioning effect is further improved.In order to verify the validity of EKF fusion algorithm under different state constraints.Firstly,the algorithm simulation results show that the pose estimation error of the EKF algorithm is less than that of the EKF algorithm under the linear equivalent state constraint,and the optimization result is better.Next,we compute attitude algorithm in terms of inertial sensor and build an experimental platform for experimental analysis.The experimental results indicate that the EKF algorithm has a good opti-mization effect on GPS fusion and inertial navigation,which the EKF algorithm with nonlinear equivalent constraint on GPS has better fusion effect than the EKF algorith-m with linear equal constraint.