Tracking algorithm analysis for the PCL-PET fusion system

We discuss the problem of poor initial estimation of altitude of distant and low located objects in PCL-PET data fusion. We focus here on tracking algorithm showing how one can make the convergence of the estimate to the real value faster. The main result is that on having replaced extended Kalman filter with unscented Kalman filter, we are able to choose from scenarios with various initial conditions the one which gives (statistically) the best estimation of position and velocity of an object.