Comparison of the Extended and Sigma-point Kalman Filters on Inertial Sensor Bias Estimation Through Tight Integration of GPS and INS

The GPS/INS integration system estimates the navigation errors as well as internal sensor errors through the use of a Kalman filter. The tight integration processing mode uses the GPS range and range-rate measurements through nonlinear Kalman filtering. The extended Kalman filtering (EKF) has been discussed in many publications on GPS/INS integration. The recently developed sigma-point Kalman filtering has advantages over the EKF, for instance, its ability to capture the high-order nonlinear terms without requiring the computation of the Jacobian matrix. The INS sensor calibration usually operates during the INS initialization. Integration of GPS and INS adds potential in-flight alignment capability to the system. This paper concentrates on the inertial sensor error estimation via nonlinear Kalman filtering, i.e. the EKF and the SPKF. A procedure for the inertial sensor bias estimation is proposed, which is based on the observability analysis of the tight integration system. In addition, an algorithm for the coarse estimation of the inertial sensor biases is proposed. The performances of the SPKF and the EKF for the fine alignment are compared.