Robust Interactive Multiple Model for Collaborative Navigation of Unmanned Surface Vessel Swarm

This paper presents a robust interactive multiple model (IMM) method to deal with the data fusion of collaborative navigation for the unmanned surface vessel (USV) swarm. Firstly, the IMM is combined with the cubature Kalman filter to handle the nonlinear characteristics involved in the collaborative navigation model based on the strategy of measuring the relative distance and azimuth. Then, a sub-filter in the IMM framework is designed as a robust filter where the Mahalanobis distance criterion is used to detect the abnormal noise statistics. Subsequently, a model selection mechanism is established to adaptively chose the proper models to cover the actual measurement noise statistics. The proposed robust IMM method has a strong adaptability and robustness to handle the varying the measurement noise statistics, it provides an effective collaborative navigation way to enhance the performance of the USV swarm. The simulation on collaborative navigation of USV swarm has been conducted to evaluate the efficacy of proposed robust IMM. The results demonstrate that it has excellent robustness compared to the traditional CKF and CKF-IMM.