An Integrated approach to the Space Situational Awareness Problem
Technical Report,15 Feb 2013,31 Aug 2016
Texas A and M Engineering Experiment Station (TEES) College Station United States
Pagination or Media Count:
We have unified the finite Set Statistics FISST and Multi-Hypothesis Tracking MHT methodologies for multi-target tracking and developed a randomized version of FISST called RFISST that makes the implementation of the full FISSTMHT recursions computationally tractable. The DDDAS paradigm used in this method actively controls the number of likely hypotheses, pruning them based on data coming from the sensors. We developed particle-based Gaussian Mixture Filters that are immune to the curse of dimensionality particle depletion problem inherent in particle filtering. This method maps the data assimilation filtering problem into an unsupervised learning problem. Results show that the performance is comparable to competing techniques. This is a thrust that is data driven to maintain the correct posterior density and uses the DDDAS paradigm. We developed a simulation based sensor scheduling scheme that can tackle the measurement steering problem inherent in all DDDAS applications. The algorithm is a heuristic solution to the underlying partially observed Markov Decision Problem PMDP that does not suffer from the curse of Dimensionality and History inherent in such problems, which allows recovery of true optimality.