Constrained Unscented Kalman Filtering for Bearings‐Only Maneuvering Target Tracking
Hongwei Zhang, Weixin Xie
Abstract
To track the bearings-only maneuvering target tracking accurately online, the soft measurement constraints are implemented into the Unscented Kalman filtering (UKF). To deal with the soft measurement constraints, the Lasso regularization is added as the obstacle function. In doing this, the sampled sigma points can be restricted into the feasible region. To enhance the sampling efficiency, the global optimal solution is acquired by a heuristic optimizer. To smooth the outliers, the posterior distribution is approximated by a Gaussian mixture consists of the original and the modified priors with the fuzzy weighted factor. Simulated results indicate the accuracy and the computational efficiency of the proposed method.