Bearings-only TMA on manoeuvring target a Gaussian mixture approach

This paper considers the problem of manoeuvring target tracking with bearings-only measurements collected by a single source. We use the switching multiple models description where the target motion is described by a finite collection of m models, each of one corresponding to a possible evolution, and a Markov chain to model the jump between the different models. We test a new approximation scheme to avoid the exponential growth in the number of terms corresponding to the optimal filter.