Multi-sensor debris tracking

We discuss the problem of tracking numerous debris objects originating from a point source, such as fragments produced by a collision or explosion. Multiple cameras are used to provide measurements of the objects, which are then fused to estimate positions and velocities in 3D space. Estimates may then be used to predict landing locations and spread, or to infer information about the event. This presentation develops a solution to such a problem, based around several high speed cameras viewing an event from a variety of angles and distances. The quantity of debris considered is both unknown and likely to be high enough that the track initiation and data association problems are especially challenging.