Application of unscented R–T–S smoothing on INS/GPS integration system post processing for airborne earth observation

The Kalman Rauch–Tung–Striebel (R–T–S) smoother has been applied to fuse data from the inertial navigation system (INS) and global positioning system (GPS) for post processing, but its optimality heavily depends on linearity. For the case of in-flight startup, the INS/GPS integration is a nonlinear system with large initial attitude errors, the linear estimation approaches become inapplicable. In this paper, the Unscented R–T–S Smoother (URTSS) is utilized to deal with the nonlinear problem in the INS/GPS integration post processing, and the performance of this algorithm is compared with a similar smoother based on Extended Kalman Filter (ERTSS) through the Monto Carlo simulations. Furthermore, an INS/GPS integration system is implemented using URTSS and applied to airborne digital camera imaging. Through numerical simulation and flight test, it is shown that URTSS has obvious accuracy advantage over ERTSS in attitude estimation.

► The state estimation problem of INS/GPS integration in-flight startup was studied.
► A possible smoothing estimation to the nonlinear problem was brought out.
► Estimate errors of different cases were compared and analyzed.
► Simulation and flight test results indicated the effectiveness of the method.