AFRL-AFOSR-CL-TR-2017-0005 Space Object Detection & Tracking Within a Finite Set Statistics Framework

This report provides a summary of the work carried out in the second year of the SOARD project, Grant No. FA9550-15-1-0069, devoted to the investigation and improvement of the detection and tracking methods of inactive Resident Space Objects (RSOs). In the second year, a Random Finite Set (RFS) based Joint Target Detection and Tracking filter was evaluated for the space object tracking scenarios and two extensions were developed in order to increase their robustness to unknown detection statistics. The performance of the extensions was evaluated using both simulated and real data from the Chilbolton Advanced Meteorological Radar (CAMRa). Both pre-processed and raw data sets from the CAMRa were obtained. In the preprocessed data set, a maximum of one detection was reported per bearing angle based on detection methods used at the CAMRa site. The raw data was also processed by using a Constant False Alarm Rate algorithm that was able to report multiple detections per bearing angle, increasing the probability of detecting the target of interest. Tracking algorithms, based on the RFS-based Joint Target Detection and Tracking (JoTT) filter were investigated, which also include the estimation of target detection statistics, parameters which are often difficult to determine, and which can be time varying. In all cases, qualitative and quantitative improvements in the robustness of the proposed tracking methods were observed when comparing with the standard JoTT filter. In addition, image sequences from the Georgia Tech observatory, known as the Omnidirectional Space Situational Awareness (OmniSSA) data set, were processed to determine the feasibility of applying the RFS filtering concepts to image data.