Two-Stage Cubature Kalman Filter for Nonlinear System with Random Bias

In contrast with UKF, the standard CKF can solve high-dimensional nonlinear filter problems. However, when the nonlinear systematic dimension increases, the accuracy of CKF will decline and the computational cost will increase rapidlly. Two-Stage Kalman filter can solve this problem, but it only applies to linear systems. This paper proposes a two-stage Cubature Kalman filter (TSCKF) which can solve high-dimensional nonlinear systems with random bias. The estimate of the TSCKF can be expressed as the output of the bias free filter and bias filter. The bias free filter doesn't consider the bias and its output is corrected by the bias filter. In comparison with Augmented state Cubature Kalman Filter (ASCKF), the TSCKF avoids dimension disaster effectively and has little computation to solve high-dimensional nonlinear filter problem.