초록 ▼

The tracking of moving objects on land, in the air or at sea is effected by means of one or more sensors (Sj1). The observation space of each sensor is divided into resolution cells (X1h) forming a grid. The sensors of the same grid are grouped together. A probability estimate (b1h) is produced. For

The tracking of moving objects on land, in the air or at sea is effected by means of one or more sensors (Sj1). The observation space of each sensor is divided into resolution cells (X1h) forming a grid. The sensors of the same grid are grouped together. A probability estimate (b1h) is produced. For this purpose, the starting points are signals (Mj1h) delivered by the sensors (Sj1) and previously selected according to windowing criteria, sets of pairs (Fi(Mj1h)Dij1h) stored in memories (1) and coming from a prior supervised statistical learning, and tracking coefficients (a4) of the PDAF type. After application of this probability (b4), there is obtained on the one hand an estimated status (X*,P*) affording a trajectory prediction (x,S) for at least one moving object and, on the other hand, the windowing criterion is adjusted.

대표청구항 ▼

A method for tracking the trajectory of moving objects, of the type in which: a) one or more sensors (Sj1) are provided, which deliver signals (Mj1h) relating to the presence of at least one moving object, with reference to a known division of space, peculiar to each group of sensors, in so-called s

A method for tracking the trajectory of moving objects, of the type in which: a) one or more sensors (Sj1) are provided, which deliver signals (Mj1h) relating to the presence of at least one moving object, with reference to a known division of space, peculiar to each group of sensors, in so-called spatial resolution cells (X1h); b) by means of logic tests peculiar to each sensor, some of these signals (Mj1h) are selected in accordance with a chosen windowing criterion, relating to the resolution cells (X1h); c) an estimate is worked out of the probability (b1h) peculiar to each sensor (Sj1), according to the signals selected at b), and trajectory tracking data (a1) the prior storage in memory of n sets of J distribution functions Fi(Mj1h), these n sets relating respectively to n identities Hi of indexed moving objects (i=1 to n; j=1 to J, each of these functions (Fi(Mj1h) representing the a priori probability (P(Mj1h/Hi)) of observing the measurement (Mj1h) in the resolution cell (X1h) of the sensor (Sj1) when a moving object of identity (Hi) is presented to this sensor, and having a coefficient (Dij1h) representing the degree of confidence given thereto; c2) the determination, from each measurement (Mj1h) and the pairs (Fi(Mj1h), Dij1h) stored in memories, of orders of likelihood (Aij1h, Bij1h), and this for all the identities (Hi) indexed, where Aij1h represents the likelihood ratio between the cases “there is a moving object of identity Hi in cell X1h”and “no moving object of identity Hi is present in the cell X1h”, whilst Bij1h represents an uncertainty factor for Aij1h; c3) the merging, for each resolution cell (X1h) and each identity of moving objects (Hi), of these pairs of orders of likelihood (Aij1h, Bij1h) obtained for all the sensors (Sj1) in the same alignment group, into a single pair of likelihood (Ai1h, Bi1h); c4) the calculation, from the likelihoods (Ai1h, Bi1h) relating to a first resolution cell (X1h), of the likelihood (Q1h) for there being in this cell (X1h) a moving object of the same identity as the moving object being tracked; c5) the calculation, from the likelihoods (Q1h) of each group of sensors, for each cell (Xk), the intersection of the resolution cells (X1h) in question, of the likelihood (Qk) of there being in this box (Xk) a moving object of the same identity as the moving object being tracked; and c6) the calculation, from the values (Qk) and tracking coefficients (aa), for each cell (Xk), of the a posteriori probability (b