This work studied two sub-problems of the cooperative state estimation and cooperative optimization of tracking paths in multiple unmanned underwater vehicle (multi-UUV) cooperative target tracking. The mathematical model of each component of the multi-UUV cooperative target tracking system was established. According to the target bearing-only information obtained by each unmanned underwater vehicle’s (UUV) detection, the extended Kalman filter algorithm based on interacting with multiple model bearing-only data was used to estimate the target state in a distributed way, and the federal fusion algorithm was used to fuse the estimated results of each UUV. The fused target state was predicted, and, based on the predicted target state, to achieve the persistent tracking of the target, the particle swarm optimization algorithm was used for the online collaborative optimization of the UUV tracking path. The simulation results showed that the multi-UUV distributed fusion filtering algorithm could obtain a better target state estimation effect, and the online path collaborative optimization method based on the prediction of the target state could achieve persistent target tracking.
25 December 2024 in IEEE Transactions on Instrumentation and Measurement
In response to the problems encountered during underwater bearing-only target tracking, namely, the increase in the actual measurement error of the flank array as the target’s relative azimuth deviates from the UUV abeam direction and the poor tracking performance due to the end-fire blind spot of the flank array, a path optimization method for UUV tracking of maneuvering targets that is specifically oriented toward achieving the best observation by the flank array is designed. A system model is established, and the variation in measurement accuracy with respect to the relative azimuth of the target is analyzed and modeled. The maneuvering target is detected when the sonar accuracy changes in real time. According to the bearing-only information measured, the target state is estimated by matching the measurement noise variance with the sonar measurement error in real time. On the basis of the estimated state, the best observation method is used to generate the maneuvering trajectory of the UUV online to reduce measurement noise during target detection. Constraints related to the end-fire blind spot are also considered to avoid situations where the target becomes unobservable. The simulation results show that the proposed method can achieve persistent target tracking while considering the actual characteristics of the flank array.
9 June 2026 in IEEE Transactions on Automation Science and Engineering
Bearing-only target tracking poses significant challenges due to the highly nonlinear nature of the measurements. To address this issue, this paper proposes a novel integrated-correction high-order extended Kalman filter (IC-HEKF). Unlike the conventional extended Kalman filter, the IC‑HEKF compensates the linearized measurement matrix through higher‑order approximations and integrates a correction step within the traditional prediction‑update framework. By relinearizing the bearing-only measurement model based on the corrected state, it overcomes the limitation of conventional methods that rely heavily on the quality of prior target information when using the one-step predicted state for linearization. As a result, the proposed method theoretically enhances the handling of nonlinearity in bearing-only measurement models and ultimately improves tracking performance. Numerical simulations demonstrate that, in a bearing‑only target tracking system, the proposed IC-HEKF achieves superior tracking performance compared to traditional methods. This improvement becomes even more pronounced under conditions where target prior information and sensor accuracy are more degraded. Finally, real‑world experiments further validate the superior performance of the IC‑HEKF.