Unscented 변환에 기반한 INS/GPS 강결합 알고리즘

Abstract

In this paper, the tightly coupled INS/GPS integration system based on unscented transform is presented for improving the performance of nonlinear kalman filter. The unscented transform is a method for predicting means and covariances in nonlinear systems and utilizes a set of samples which are deterministically chosen. These samples are scaled to guarantee the mean and covariance second order accuracy by an arbitrary constant. We introduce the scaled unscented transformation which modifies the sigma points themselves rather than the nonlinear transformation. And then, the scaled method is used to transform the pseudo range measurement of the tightly coupled approach. To analyze the performance of the UKF based tightly INS/GPS integration system, real van test and simulation have been carried out. The results show that the UKF and EKF based approaches offer an identical performance except for the presence of the unmodeled noise.