Gaussian Mixture Unscented Particle Filter with Adaptive Residual Resample for Nonlinear Model
Na Zhang, Xin Yang · 2015
To solve nonlinear non-Gaussian filter problems in target tracking, Gaussian mixture unscented particle filter with adaptive residual resample algorithm is proposed.Gaussian mixture unscented particle filter is utilized as importance density to improve the estimation accuracy evidently.By introducing adaptive residual resample, the new algorithm overcomes the defects of general resample algorithm.To evaluate the proposed algorithm, the random surfer dynamic model and range-rate measurement are involved as nonlinear models with two static sensors.Simulation results show that the proposed algorithm performs robust and effective.As a consequence, compared with the general Gaussian particle filter, the proposed algorithm is more accurate in estimated state and more diverse in particles.