Radar Debris Algorithms and Models for Discrimination

We are proposing an innovative approach to sensor fusion and target recognition based on transductive inference with support vector machines. Our proposed algorithm will apply innovative discrimination to the fusion of sensor (feature) and contextual scenario information through the development of robust algorithms and software necessary to collect, process, and fuse information from multiple sources (radars either at the same or different frequencies as well as EO/IR sensor assets). This proposed approach has the advantage of being able to combine these disparate data types of high dimensionality and use small training sets to achieve unprecedented performance. The goal of the data fusion process described in this proposal is to operate on a combination of sensor measurements, features, track states, and object type and identification likelihoods to produce a highly accurate integrated picture of the battlespace. Our results will be compared with Professor Donald Maurer's work at Johns Hopkins Applied Physics Laboratory, where he conducted research on using Bayesian network for the same task.