Multiple sliding surface guidance for planetary landing : tuning and optimization via reinforcement learning

Wibben, Daniel and Gaudet, Brian and Furfaro, Roberto and Simo, Jules; Tanygin, Sergei and Park, Ryan S. and Starchville Jnr., Thomas F. and Newman, Lauri K., eds. (2013) Multiple sliding surface guidance for planetary landing : tuning and optimization via reinforcement learning. In: Advances in the Astronautical Sciences. Advances in the Astronautical Sciences, 148 . Univelt Inc, USA, pp. 1881-1900. ISBN 9780877035978

[thumbnail of Simo J et al - AAM-Multiple-sliding-guidance-for-planetary-landing-Tuning-and-optimizing-via-reinforcement-learning-2013] PDF. Filename: Simo_J_et_al_AAM_Multiple_sliding_guidance_for_planetary_landing_Tuning_and_optimizing_via_reinforcement_learning_2013.pdf
Accepted Author Manuscript
Download (1MB)

Abstract

The problem of achieving pinpoint landing accuracy in future space missions to extra-terrestrial bodies such as the Moon or Mars presents many challenges, including the requirements of higher accuracy and more flexibility. These new challenges may require the development of novel and more advanced guidance algorithms. Conventional guidance schemes, which generally require a combination of off-line trajectory generation and real-time, trajectory tracking algorithms, have worked well in the past but may not satisfy the more stringent and difficult landing requirements imposed by future mission architectures to bring landers very near to specified locations. In this paper, a novel non-linear guidance algorithm for planetary landing is proposed and analyzed. Based on Higher-Order Sliding Control (HOSC) theory, the Multiple Sliding Surface Guidance (MSSG) algorithms has been specifically designed to take advantage of the ability of the system to reach the sliding surface in a finite time. The high control activity seen in typical sliding controllers is avoided in this formulation, resulting in a guidance law that is both globally stable and robust against unknown, but bounded perturbations. The proposed MSSG does not require any off-line trajectory generation and therefore it is flexible enough to target a large variety of point on the planet’s surface without the need for calculation of multiple reference trajectories. However, after initial analysis, it has been seen that the performance of MSSG is very sensitive to the choice in guidance gains. MSSG generated trajectories have been compared to an optimal solution to begin an investigation of the relationship between the optimality and performance of MSSG and the selection of the guidance parameters. A full study has been performed to investigate and tune the parameters of MSSG utilizing reinforcement learning in order to truly optimize the performance of the MSSG algorithm. Results show that the MSSG algorithm can indeed generate trajectories that come very close to the optimal solution in terms of fuel usage. A full comparison of the trajectories is included, as well as a further study examining the capability of the MSSG algorithm under perturbed conditions using the optimized set of parameters.

CORE (COnnecting REpositories)