Research on Collaborative Navigation Algorithm of Factor Graph Based on BP

The standard particle filter (PF) algorithm has the problem of particle diversity loss caused by particle impoverishment and resampling, which makes the particle sample unable to accurately represent the true distribution of the state probability density function. In this paper, a particle sampling method based on Sigma points and a method of classifying particles and recalculating weights based on weights are proposed. Based on the improvement of these two parts, an improved BP cooperative navigation algorithm is proposed. The simulation results show that the navigation error estimated by the improved BP algorithm proposed in this chapter is smaller than that of the traditional NBP algorithm, and the effectiveness of the improved algorithm is verified.