• three-body problem
• SDRE
• imitation learning
• neural networks
• rendezvous

The aim of this thesis it to analyze how to design a closed loop control based on neural networks for a rendezvous in a three body problem scenario considering both the translational and attitude dynamics. The rendezvous is between two spacecraft, the chaser which is actively controlled and the target wich is uncooperative and it is located on a Near Rectilinear Halo Orbit (NRHO). The guidance system synthesis is done by means of an SDRE control and a PID control. Then, based on the idea of ‘Imitation Learning’, 2 neural networks, one for the transaltional dynamics and one for the attitude dynamics, have been trained in each case to learn the SDRE and PID control behaviours, respectively. After the training, they have been inserted into the closed-loop dynamical system to asses their capabilities to 'imitate' the control behaviours that they have learnt.