ESA GNC Conference Papers Repository

Future space missions plan to tackle problems such as on-orbit inspection, servicing and assembly, and debris removal. Inherently, these missions involve a much higher degree of uncertainty due to the unknown state of the targets and the fact that many of these targets were not designed with docking capabilities (sensors, docking ports, etc.) in mind. In order to design proper controllers and estimators, it is imperative to understand the effects of uncertainties on design level decisions. Choosing the appropriate architecture for a particular mission phase will depend both on the mission objective and the level of uncertainty in key system parameters. The purpose of this paper will be to analyze the performance of adaptive versus static controllers and Bayesian inference versus Kalman filtering. Of interest are performance metrics such as robustness and stability margins, propellant consumption, tracking error and maneuver completion time. Several techniques will be compared such that the best performing control and estimation architecture is chosen for a particular mission phase and uncertainty level. Uncertain parameters under consideration include unknown mass properties, actuator performance, target orbit properties and orbital perturbations. In addition, a trade on the merit of performing system identification after docking to an uncertain target or using adaptive techniques to overcome the parameter uncertainty will be explored. The study will present results from a simulator based on Monte Carlo methods and will derive the probabilistically most effective controller and estimator schedule for the multi-stage proximity operation mission profile provided. Additionally, results will be provided to show which techniques are suitable across a range of uncertainty levels.