AN AEGIS-FISST ALGORITHM FOR JOINT DETECTION, CLASSIFICATION AND TRACKING

Space Situational Awareness requires the acquisition, cataloging, retrieval and predictive use of a large amount of data pertaining to the existence, character, and orbital state of all objects in Earth orbit. About five hundred thousand objects one centimeter in size or larger populate the space around earth, which increases the need for new methods and techniques that can contribute to the core tasks of SSA: discovery of new objects, tracking of (possibly maneuverable) detected objects, and classification of tracked objects. A key feature of the SSA problem is the fact that these tasks are interdependent. This interdependence between hybrid continuous and discrete estimation variables makes the joint detection, classification and tracking inference problem (JDCT) coupled. One method that takes into account such interdependencies is the Finite Set Statistical (FISST) hybrid multi-target estimation approach and its derivative methods. In this paper we derive the multi-target multi-class FISST (MTMC FISST) filter as well as a Gaussian mixture-based closed-form approximation to the filter, resulting in the MTMC AEGIS-FISST filter. As a proof of concept, we demonstrate the main result using a simple single-object, dual-class SSA problem.