ACTA and Sandia National Laboratories will develop and demonstrate advanced concepts for processing large and disparate data streams from multiple classes of sensors using multi-physics based mappings based upon stochastic non-linear networks. Fusion of incertitude will be demonstrated with algorithms based on KDE/ICA (kernel density estimator/independent component analysis) approaches which yield truncation-free and viable non-parametric applications of polynomial chaos expansions. ACTA is prepared to incorporate these multi-physics mapping algorithms into its existing embeddable sensor-fusion hardware design constructed around a tool called REASON to produce a working prototype of what it currently terms as Smart Remotes. REASON stands for Rule Extensible Arbitrary System of Networks, which was developed in part to avoid the pitfalls of monolithic knowledge-based systems, and also to incorporate"in-line"as much as possible modern computationally-oriented syntaxes and to utilize open-source data structures and storage formats.
Benefit: Air Force mission readiness, tactical situation awareness, and weather forecasting are areas where the proliferation of sensor platforms can obviously augment existing capabilities, increase decision turn-around, and amplify or extend solider and pilot effectiveness. These goals are not achievable however without adequate algorithms to map the multi-physics suite of data into feature discrimination and change point detection estimates with corresponding joint densities in order to operate with acceptable false alarms and detection performance. It is vitally important that the multi-physics algorithms be based on a sound theoretical underpinning, and be able to be deployed across a wide variety of platforms. ACTA"s proposed technical approach will demonstrate that these benefits can be achieved.
Keywords: sensor fusion, multi-physics mappings, thermomagnetoelastic models, joint density estimation, Independent Components Analysis, kernel density estimators, polynomial chaos expa
