Robust iterated q-Rênyi unscented Kalman filter for spacecraft attitude estimation

Since the process noise of the spacecraft kinematic model has state coupling properties, the complex external space environment influences the star sensor. These factors undermine the Gaussian noise hypothesis and are likely to result in non-Gaussian noise characteristics. The traditional Kalman filter, derived under the assumption that the noise is Gaussian distributed, cannot satisfy the system’s accuracy requirements under non-Gaussian noise situations. An iterative unscented Kalman filter based on the q-Rênyi kernel function is proposed in this paper. The augmented model combines state error and measurement error in the cost function. The propagation equations are utilized to give the priori estimates, and the new fix-point algorithm is utilized to give the posteriori estimates. In order to avoid the Cholesky decomposition process of non-positive definite symmetric matrices, hyperbolic QR decomposition is introduced in this paper for optimization. Finally, the effectiveness of the proposed algorithm is verified under non-Gaussian noise environments.