Cooperative Target State Estimation of Multiple AUVs Based on an Enhanced IMM-UKF Approach

To improve the accuracy of cooperative target state estimation for autonomous underwater vehicles (AUVs) under an unknown target motion model, this paper presents an estimation framework based on the interacting multiple model unscented Kalman filter (IMM-UKF). The proposed method integrates three tailored components to enhance estimation performance under bearing-only observations: (1) a least-squares cross-location initialization strategy to improve the filter's convergence under nonlinear measurements; (2) an adaptive model probability update mechanism that incorporates inter-AUV residual information to improve motion model discrimination; and (3) an information-matrix-weighted fusion approach that accounts for the varying confidence levels of individual AUV estimates. The results show that the proposed method can accurately estimate the target's motion state and significantly improve the state estimation accuracy and robustness in multi-AUV cooperative observation, which provides an effective technical solution for multi-platform cooperative sensing in the complex marine environment.