2. Estimation principles and algorithms for use in practical work

2.1 Choosing a state vector

Before embarking on the problem of estimating the state vector, it is advisable to check its observability, which means knowing how many components of the state vector can be determined. For example, in TP2A (passive trajectography by measuring two azimuths), the state vector composed of two position coordinates and two velocity coordinates (so dimension d = 4) is observable, unless the noisemaker moves on the straight line passing through the two locations where the two sensors are located. Conversely, in TPA, if the listener is not moving, it is illusory to try to estimate this same state vector: we show