Trajectory Prediction Algorithm Based on Variational Bayes

This paper bring forward a variational Bayes trajectory prediction algorithm based on Gaussian mixture trajectory model. It is assumed that the movement patterns of moving target can be statistically modeled by a common Gaussian mixture model, and this algorithm contains three main steps: 1.Data preprocessing. 2. Ttajectory model training. 3. Trajectory prediction. Meanwhile, this paper analysed the error of this algorithm, according to the real-world data set, the simulation result indicates that: Compared with EM prediction algorithm under the same parameter setting, the trajectory prediction accuracy of VB prediction algorithm is increased by 87.3% on average.