Adaptive interacting multiple model for underwater maneuvering target tracking with one-step randomly delayed measurements

The existing interacting multiple model (IMM) suffers from two problems when tracking an underwater maneuvering target: 1) the conventional nonlinear filters in IMM cannot work on the condition of delayed measurement, 2) the inefficiency of model switching in IMM may degrade the tracking performance. In order to handle the delayed measurement, a novel Bayesian filter framework under two Gaussian assumptions is proposed. The new extended Kalman filter (EKF) and unscented Kalman filter (UKF) are developed as two variations of the proposed filter. In addition, to solve the problem of low switching efficiency between IMM models, a transfer probability corrected function is designed to adaptively update the transfer probability matrix (TPM) in IMM. The simulation results show that the proposed algorithm outperforms the existing IMM-EKF and IMM-UKF.