Rocket reusability is a key factor in enabling quicker and more cost-effective access to space. However, landing on Earth poses significant challenges due to the dynamic and highly uncertain environment. A robust Guidance, Navigation, and Control system is essential to guide the vehicle to the landing site while meeting terminal constraints and minimizing fuel consumption. This research integrates Meta-Reinforcement Learning with Gated Transformer XL Neural Networks to enhance the robustness of the powered guidance with respect to atmospheric and aerodynamic uncertainties, navigation and control errors, and dispersed initial conditions. By employing a 6-Degrees-of-Freedom dynamics model and accurate vehicle and environmental simulations, the agent learns a higher fidelity guidance policy compared to existing literature, demonstrating successful and robust performance in Monte Carlo simulations. In this complex scenario, the innovative attention-based neural networks also outperform recurrent neural networks, widely used for Reinforcement Learning-based space guidance applications.

In this paper a new drag-energy scheme for atmospheric entry guidance, based on the use of pseudospectral methods and convex optimization, is proposed. One of the most successful technologies to deal with atmospheric entry is the class of drag-tracking schemes, a direct heritage of the Space Shuttle program. The method that we propose exploits the drag-dynamics, and allows for an efficient automatic design of an optimal entry profile satisfying all the longitudinal constraints acting on the vehicle. A new representation of the entry-guidance problem, able to loss-less convexify the formulation, is provided. Numerical simulations confirm the validity of the proposed scheme as tool for further improving the autonomy of modern entry guidance systems, with a mean final range-to-go error at the end of entry phase smaller than 2 km, and the capability to re-compute a complete constrained trajectory to meet the mission requirements.

In this paper a new drag-energy scheme, based on the use of pseudospectral methods and convex optimization, is proposed. One of the most successful technologies to deal with atmospheric entry is the class of drag-tracking schemes, a direct heritage of the Space Shuttle program. The method that we propose exploits the drag-dynamics, and allows for an effcient automatic design of an optimal entry profile satisfying all the longitudinal constraints acting on the vehicle. A new representation of the entry-guidance problem, able to loss-less convexify the formulation, is provided. Numerical simulations confirm the validity of the proposed scheme as tool for further improving the autonomy of modern entry guidance systems, with a mean final range-to-go error in the order of three hundred meters, and the capability to re-compute a complete constrained trajectory to meet the mission requirements.