Technology for Low Resolution Space Based RSO Detection and Characterisation

Space Situational Awareness (SSA) refers to all activities to detect, identify and track objects in Earth orbit. SSA is critical to all current and future space activities and protect space assets by providing access control, conjunction warnings, and monitoring status of active satellites. Currently SSA methods and infrastructure are not sufficient to account for the proliferations of space debris. In response to the need for better SSA there has been many different areas of research looking to improve SSA most of the requiring dedicated ground or space-based infrastructure. In this thesis, a novel approach for the characterisation of RSO’s (Resident Space Objects) from passive low-resolution space-based sensors is presented with all the background work performed to enable this novel method. Low resolution space-based sensors are common on current satellites, with many of these sensors being in space using them passively to detect RSO’s can greatly augment SSA with out expensive infrastructure or long lead times. One of the largest hurtles to overcome with research in the area has to do with the lack of publicly available labelled data to test and confirm results with. To overcome this hurtle a simulation software, ORBITALS, was created. To verify and validate the ORBITALS simulator it was compared with the Fast Auroral Imager images, which is one of the only publicly available low-resolution space-based images found with auxiliary data. During the development of the ORBITALS simulator it was found that the generation of these simulated images are computationally intensive when propagating the entire space catalog. To overcome this an upgrade of the currently used propagation method, Specialised General Perturbation Method 4th order (SGP4), was performed to allow the algorithm to run in parallel reducing the computational time required to propagate entire catalogs of RSO’s. From the results it was found that the standard facet model with a particle swarm optimisation performed the best estimating an RSO’s attitude with a 0.66 degree RMSE accuracy across a sequence, and ~1% MAPE accuracy for the optical properties. This accomplished this thesis goal of demonstrating the feasibility of low-resolution passive RSO characterisation from space-based platforms in a simulated environment.