Application of Sigma-Point Filtering to GPS/INS Integrated Navigation System with Simulation

A new fusion algorithm of the measurements from global positioning system(GPS) and inertial navigation system(INS)is presented to compensate the weaknesses of current integrated navigation system. The extended Kalman filtering (EKF), an algorithm commonly used in integrated navigation system is a suboptimal one in a sense because it simplifies the nonlinear dynamic system model to a first order linear model and assumes that the linearized one is driven by Gaussian white noise. This simplification leads to questionable results in certain scenarios. One of the improved versions of EKF is sigma-point Kalman filtering (SPKF). SPKF needn′t assume a linearized dynamic model, and uses samples set to represent the state distribution, leading itself an adequate candidate for navigation applications. One of the good features of SPKF is that the states′ mean and covariance can be easily obtained. For Gaussian random variables, the mean and covariance can accurately approximate to at least second-order in any nonlinear system. Simulation results show that SPKF works well in GPS/INS integrated navigation system.