Adaptive generalized ZEM-ZEV feedback guidance for planetary landing via a deep reinforcement learning approach. (June 2020)

• Record Type: Journal Article
• Authors: Furfaro, Roberto; Scorsoglio, Andrea; Linares, Richard; Massari, Mauro
• Is Part Of: Acta astronautica. Volume 171(2020)
• Page Start: 156
• Page End: 171
• Publication Date: 2020-06
• Subjects: Optimal landing guidance -- Deep reinfocement learning -- Closed-loop guidance; Astronautics -- Periodicals; Outer space -- Exploration -- Periodicals; Astronautics; Periodicals; 629.405
• DOI: 10.1016/j.actaastro.2020.02.051
• Languages: English
• ISSNs: 0094-5765

Abstract

Abstract: Precision landing on large and small planetary bodies is a technology of utmost importance for future human and robotic exploration of the solar system. In this context, the Zero-Effort-Miss/Zero-Effort-Velocity (ZEM/ZEV) feedback guidance algorithm has been studied extensively and is still a field of active research. The algorithm, although powerful in terms of accuracy and ease of implementation, has some limitations. Therefore with this paper we present an adaptive guidance algorithm based on classical ZEM/ZEV in which machine learning is used to overcome its limitations and create a closed loop guidance algorithm that is sufficiently lightweight to be implemented on board spacecraft and flexible enough to be able to adapt to the given constraint scenario. The adopted methodology is an actor-critic reinforcement learning algorithm that learns the parameters of the above-mentioned guidance architecture according to the given problem constraints. Highlights: A Deep Reinforcement Learning (DRL) framework for structured planetary landing guidance is presented. The DRL framework is based on a customized actor-critic approach that employs Extreme Learning Machines (ELM). The algorithm learns the ZEM/ZEV generalized feedback guidance gains as function of the spacecraft state. The gains are adapted to ensure both quasi-optimality and flight-constraints. Stability analysis shows the proposed approach is globally stable.