Variational Measurement Update for Extended Object Tracking Using Gaussian Processes

We present an alternative inference framework for the Gaussian process-based extended object tracking (GPEOT) models. The method provides approximate solution to Bayesian filtering problem in GPEOT by relying on a new measurement update, which we derive using variational Bayes techniques. resulting algorithm effectively computes posterior densities of kinematic and extent states. conduct various experiments simulated real data examine performance compared with reference method, employs Kalman filter inference. proposed significantly improves accuracy both estimates proves robust against model uncertainties.