Ryan Clark, York University; Siddharth Dave, York University; Regina Lee, York University
Keywords: Space Situational Awareness, Resident Space Object, Star Trackers, Parameter Optimization
Abstract:
The lack of Space Situational Awareness (SSA) for Low Earth Orbit has been a growing concern in the space sector. With the number of tracked objects being estimated to double in the next decade, better SSA methods are required to keep up with this increase. Optical detections have been demonstrated to be able to estimate a Resident Space Object (RSO) shape, attitude and optical properties with a sufficiently accurate priory information. This is accomplished by inverting an RSOs light curve, where a RSOs light curve is the received flux, usually in the form of brightness, vs time. A RSOs brightness is a complex function that takes into account the solar phase angle, RSOs shape, attitude, and reflectance. Different light curve inversions techniques have demonstrated feasibility from ground based optical detections but has not been demonstrated on passive space based optical sensors such as Star Trackers. Space based detectors have some advantages over ground based detectors including not having to view through the atmosphere, getting different viewing geometries for Earth facing satellites, and being able to take observations anytime of the day. This paper presents the results of a feasibility study for a novel application of estimating an RSOs attitude from space based passive detections. For the first part of the feasibility study a star tracker image simulator was used to simulate space based RSO detections and produce light curves from RSOs in a controlled environment. The simulation uses a forward ray tracing model to recreate the image. Four different RSO shape models will be used; 1U CubeSat, 3U CubeSat, Box-Wing Cube Satellite, and Iridium First Generation models. To model the light reflected off of the models are broken down in to 2D fixed shape facets, where the directional reflectance distribution function is well known. The 3 facets that are used to in this model are: flat surface, cylindrical and spherical. Once this novel application has been demonstrated on simulated data the second part of the feasibility study is to confirm the results with star tracker like space based images. In this feasibility study five different methods of parameter optimization for an RSOs shape, attitude and optical parameters are being tested: Gradient Descent, Stochastic Hill Climbing, Simulated Annealing, a Genetic Algorithm and Particle Swarm Optimization. The feasibility of this novel application will be based off of the accuracy of the estimated parameters, as well as, the computational efficiency of each algorithm to converge on a RSOs attitude. The end goal of this research is to demonstrate the feasibility of using in-situ star trackers to passively characterise the physical properties of an RSO, which would allow it to augment current SSA information. Future research in this area will involve developing flight ready hardware to enabling this parameter estimation to happen near real time. Generations and curation of space based RSO star tracker detections to serve as data sets for method validation and to train machine learning algorithms. Better prediction models for space based RSO detections. Orbital parameter estimation from space based optical observations.
Date of Conference: September 15-18, 2020
Track: Non-Resolved Object Characterization