Optimal trajectory planning of a satellite for in-orbit experiments on ISS

Abstract

Optimal trajectory planning is considered as a solved problem in robotics for offline situations. On the other hand, for real time operation of an implemented solution on a physical robot, the computational time for the task execution may still be prohibitive. The goal of this work, developed at the Deutsches Zentrum für Luft- und Raumfahrt(DLR), was to implement procedures that could be used in a space mission, considering also time as a resource. For this we developed two methodologies in order to shift the heavy part of the computation from an online to an offline context. The first approach uses a grid in the task space to use as Look Up table for generating initial guess for a lighter online optimization, while the second approach provides an initial guess for the online optimization through Multivariate Regression. Results are presented for a satellite maneuver in an in-orbit experiment planned on the ISS.