Discrete-time invariant extended Kalman filter on matrix Lie groups

Abstract

In this article, we derive symmetry preserving discrete-time invariant extended Kalman filters (IEKF) on matrix Lie groups. These Kalman filters offer an advantage over classical extended Kalman filters as the error dynamics for such filters are independent of the group configuration which, in turn, provides a uniform estimate of the region of convergence. In contrast to existing techniques in the literature, the discrete-time IEKF is derived using minimal tools from differential geometry which simplifies the derivation and the representation of IEKF. In our technique, the linearized error dynamics is defined on the Lie algebra directly using variational approaches, unlike conventional approaches where the error dynamics is translated to an Euclidean space using the logarithm map before its linearization. Moreover, the Kalman gains and its associated difference Riccati equations are derived in operator spaces by setting a discrete-time optimal control problem and solving it with discrete-time Pontryagin's maximum principle. The proposed discrete-time IEKF is implemented for the attitude dynamics of the rigid body, which is a benchmark problem in control. It is observed from the numerical studies that the IEKF is computationally less intensive and provides better performance than the classical extended Kalman filter.