Multiplicative quaternion extended consensus Kalman filter for attitude and augmented state estimation

Abstract

This paper provides a novel algorithm for the estimation of spacecraft attitude and augmented parameters (the angular velocity and the gyroscope drift). We use the quaternion to represent the attitude of a rigid body and the attitude kinematics and dynamics of the spacecraft are naturally described by nonlinear equations. Multiplicative extended Kalman filter (MEKF) is a classical algorithm to estimate the quaternion from spacecraft attitude kinematics and dynamics. In recent years, advanced spacecraft missions are typically equipped with multiple intelligent nodes, where each node has independent sensing and processing abilities embedded with a suite of sensors including star trackers and gyroscopes. Therefore, these intelligent nodes can be considered as an aggregated multi-agent system. The information obtained from the measurements of sensing nodes is exchanged through communications among different nodes. Under this framework, we propose a novel algorithm, i.e., multiplicative extended Kalman consensus filter (MEKCF) to take advantage of information sharing mechanism. In this algorithm, each node operates a local MEKF estimator and updates its state using both its own information and information from its neighbors. In particular, the estimated states for each spacecraft are the quaternion, angular velocity, and the drift parameter of each node's gyroscope. Simulations are provided to verify the validity of the proposed algorithm, where a set of nodes running MEKCF are compared with an equivalent set of isolated local nodes running MEKF. The comparison results demonstrate that MEKCF presents a faster convergence velocity and better steady estimation precision.