Attitude fusion approach for multisensor with unknown noise characteristics

A covariance intersection fusion estimation algorithm is presented for multi-sensor system with unknown covariance and noise characteristics, which means that the relativity and variance of the state noise and measurement noise are not available. Firstly, a corresponding CKF estimator is chosen for each subsystem to produce a local estimation according to the measurement data newly acquired. Secondly, based on the least matrix-weighted linear variance rule, a fast successive covariance intersection (SCI) fusion algorithm is proposed to achieve the optimal fusion estimation, which simplifies the multidimensional optimization problem into the optimization of several one-dimensional nonlinear cost functions. Thirdly, in the subsystems, a method of switching between error quaternion and error modified rodrigues parameter is used to avoid the norm constraint of quaternion and the singular problem of modified rodrigues parameter. Finally, numerical simulations are presented to demonstrate the effectiveness of the proposed algorithm.