Computational Framework for Target Tracking Information Fusion Problems

In this work, we propose computationally tractable techniques for extracting valuable information from diverse data sources collected by multiple sensors in a variety of formats (visual, sonar, quantitative, qualitative, social information, etc.). More specifically, we develop an integrated approach consisting of two algorithms for extracting information and achieving a consensus-based, robust solution. The first algorithm extracts solutions from sensors within each data source, whereas the second algorithm reaches a compromise among the generated solutions from the previous algorithm across all data sources. To accomplish these goals, we initially transform the multisensor multitarget tracking problem (MSMTT) problem into a multidimensional assignment problem. Subsequently, we introduce a decomposition-based multisensor recursive approach referred to as a revised multisensor recursive algorithm, which can efficiently deliver a robust solution for each single data source MSMTT problem. In the second algorithm, we extend our methodology to the multisource MSMTT problem by introducing a connection-based symmetric nonnegative matrix factorization technique, which is shown to be computationally feasible and efficient in obtaining high-quality solutions. History: Accepted by Ram Ramesh, Area Editor for Data Science & Machine Learning. Funding: This work was supported by the Army Research Laboratory [Grant G00006831]. Supplemental Material: The software that supports the findings of this study is available within the paper and its Supplemental Information ( https://pubsonline.informs.org/doi/suppl/10.1287/ijoc.2023.0016 ) as well as from the IJOC GitHub software repository ( https://github.com/INFORMSJoC/2023.0016 ). The complete IJOC Software and Data Repository is available at https://informsjoc.github.io/ .