LAUSR

Abstract This paper focuses on the feasibility criterion of fuel-optimal powered landing problems. Due to the uncertainties during landing on the surface of the planet, the initial states of the powered descent and landing phase may be placed in an infeasible region. Clarifying the feasibility criterion of the fuel-optimal problem is critical to ensure the convergence of the powered landing guidance algorithm. According to Pontryagin's maximum principle, the fuel-optimal conditions for the powered landing problem and its inverse problem are derived. By analyzing the inverse problem, a data generation strategy is proposed to generate sample trajectories. Deep neural networks are used to fit the parameter correlation and construct the feasibility criterion. Numerical simulations are presented to evaluate the effectiveness of the proposed deep-neural-network-based feasibility criterion and further illustrate the feasible regions of specific scenarios.