Sequential Filtering Techniques for Simultaneous Tracking and Parameter Estimation

Abstract

The number of resident space objects is rising at an alarming rate. Mega-constellations and breakup events are proliferating in most orbital regimes, and safe navigation is becoming increasingly problematic. It is important to be able to track RSOs accurately and at an affordable computational cost. Orbital dynamics are highly nonlinear, and current operational methods assume Gaussian representations of the objects' states and employ linearizations which cease to hold true in observation-free propagation. Monte Carlo-based filters can provide a means to approximate the a posteriori probability distribution of the states more accurately by providing support in the portion of the state space which overlaps the most with the processed observations. Moreover, dynamical models are not able to capture the full extent of realistic forces experienced in the near-Earth space environment, and hence fully deterministic propagation methods may fail to achieve the desired accuracy. By modeling orbital dynamics as a stochastic system and solving it using stochastic numerical integrators, we are able to simultaneously estimate the scale of the process noise incurred by the assumed uncertainty in the system, and robustly track the state of the spacecraft. In order to find an adequate balance between accuracy and computational cost, we propose three algorithms which are capable of tracking a space object and estimating the magnitude of the system's uncertainty. The proposed filters are successfully applied to a LEO scenario, demonstrating the ability to accurately track a spacecraft state and estimate the scale of the uncertainty online, in various simulation setups.