Data driven solutions for general satellite maneuvers.(NPO-50934-1)

Satellite swarms will be an integral tool for many future space-based scientific endeavors, including space-based interferometry. These endeavors will require the ability to precisely reconfigure satellites swarms to specifications generated on the fly according to scientific priorities. Crucial to reconfiguring swarms is the ability to maneuver individual satellites between prespecified start and end positions in a fixed amount of time.However, there is no consensus solution on how to do these single satellite maneuvers fuel-efficiently in the general case. Consider the position of a single satellite in the swarms Local Vertical/Local Horizontal (LVLH) reference frame relative to the swarms reference orbit. For specific single satellite transition scenarios, such as transferring the satellite between LVLH passive reference orbits (PROs) with minimal thrust expenditure in the Clohessy-Wiltshire (CW) linearized orbit dynamics system, provably optimal algorithms have been developed. But these provably optimal algorithms fail to generalize in a multitude of ways. They fail to generalize to arbitrary start and end states in the LVLH frame, they fail to generalize to more complex dynamical models such as those including J2 and drag terms, they cannot handle more detailed models of thrusters and satellite pose, and cannot account for more complex optimization criteria which balance the needs of scientific instruments with fuel efficiency and timeliness.