Abstract
A novel method for efficient high-accuracy satellite attitude estimation is presented to address the increasing performance requirements of resource-constrained small satellites. Symplectic numerical methods are applied to the nonlinear estimation problem for Hamiltonian systems, leading to a new general solution that exactly preserves state probability density functions and conserves invariant properties of the dynamics when solving for the state estimate. This nonlinear Symplectic Filter is applied to a standard small satellite mission and simulation results demonstrate orders of magnitude improvement in state and constants of motion estimation when compared to extended and iterative Kalman methods, particularly in the presence of nonlinear dynamics and high accuracy attitude observations. Based on numerous simulations, the authors conclude that this new method shows promise for improved attitude estimation onboard high performance, resource-constrained small satellites.