This program will develop an innovative Random Finite Set (RFS)-theory-based software tool for Multi-Target Tracking (MTT), using measurement filtering methods that include the Sequential Monte Carlo Generalized Labeled Multi-Bernoulli (SMC-GLMB) and the Student’s t-Mixture GMLB (STM-GLMB) filters. These MTT methods enable classification and tracking of objects within the field of view of spacecraft, including a target spacecraft for rendezvous, secondary spacecraft, orbital debris, or other planetary bodies. In this program, ASTER Labs’ team will develop RFS-based algorithms that will improve the reliability of sensor measurement gathering, object classification, and target tracking, even in the presence of high levels of non-Gaussian noise. The newly developed RFS-MTT Toolset will integrate RFS-based algorithms with Clohessy-Wiltshire-Hill, Tschauner-Hempel, and Karlgaard relative orbital dynamics equations, sensor and uncertainty models, and non-Gaussian noise-generation methods to form a full software package for simulation and analytical purposes. Orbital trajectory data from databases maintained by NORAD that feature multiple rendezvous maneuvers will be utilized along with noise models to create additional measurement uncertainty. This data will be processed via the developed RFS-MTT Toolset to confirm fidelity of the dynamics models, analyze the RFS-based algorithms, and verify the algorithms’ ability to accurately track targets in high-clutter and high sensor noise environments. Phase I will focus on developing the RFS-MTT Toolset and associated algorithms for simulations and performance assessment in orbital spacecraft rendezvous and proximity operations. The project will also evaluate these algorithms for eventual incorporation into NASA’s existing software tools, e.g. GEONS.
This RFS-MTT Toolset will be directly applicable to NASA’s spacecraft rendezvous and proximity operations missions. The software will enhance spacecraft multi-target tracking capabilities, to detect other vehicles and objects in the presence of non-Gaussian noise and false positives. The software applies to cargo transport and delivery, satellite servicing, and orbital debris removal, which will improve modeling and performance in an increasingly cluttered space environment, while having broader applicability to aerial and ground vehicles.
The RFS Multi-Target Tracking algorithms apply to systems requiring data-driven solutions for target identification, classification, and tracking in high-noise environments. Non-NASA applications include military hostile satellite tracking, and covert operations. Commercial applications include UAS integration into civilian aerospace, integration onto UGV systems, and pedestrian flow monitoring.