Integrated guidance for Mars entry and powered descent using reinforcement learning and pseudospectral method

Abstract

Traditional studies on Mars entry, descent, and landing usually take a divide-and-conquer approach where each phase is investigated separately. This paper proposed a new approach to achieve optimal integrated guidance for mid-lift high-mass Mars entry and powered descent. First, optimal Mars atmospheric entry and optimal powered descent are respectively modeled. Second, optimal handover problem is formulated to integrate the optimal Mars entry and powered descent problems in a collaborative optimization manner. Third, hp-adaptive pseudospectral method is employed to capture optimal entry guidance and propellant-optimal powered descent guidance. Reinforcement learning methodology is adopted to obtain the optimal handover, according to a large amount of trial results obtained from solving optimal Mars entry and powered descent problems. Finally, numerical simulations verified the optimality, robustness, and accuracy of the proposed technique through comparing with successive convexification approach under the nominal conditions and uncertainties, and the comparison results demonstrated the significant benefit of the integrated optimal guidance.