Unscented particle filter with risk sensitive function

In order to overcome the poor robustness and divergence which cause by uncertain estimation model, risk sensitive estimator was inforduced into the unscented particle filter, the proposed algorithm could automatically change the state noise covariance according to the magnitude of the risk function. As a result, sample impoverishment could be mitigated, and the robustness of filter would be improved. A simulation example of submarine bearing and frequency tracking was presented, the performance of the proposed algorithm was compared with the unscented Kalman filter and the unscented particle filter. Simulation results show that the new algorithm performs better than the two others.