Research on robust unscented regularized particle filtering
Li Xue, Shesheng Gao, Jianchao Wang · 2010
In nonlinear and non-Gaussian systems, particle filtering is effective but it is difficult to select the importance distribution function and diverges more greatly. Aiming at this problem, the paper represents robust unscented regularized particle filtering to improve the performance of filtering. This algorithm is more suitable for filtering calculation in nonlinear system, not only because overcomes the limitations of the general particle filter and uses the equivalent weight, but also takes advantage of the high efficiency of unscented particle filtering and regularized particle filtering. In importance sampling process, the UT transformation is applied and the equivalent weight makes good use of more reasonable information, it considers the latest measured values and slows down the particle degradation. In resampling process, particles are from the continuous kernel density distribution function owned the minimum mean square error. Simulation results show that the algorithm is efficient and outperforms in terms of accuracy based on SINS/SAR integrated navigation system.