Neural Network-Based Self-triggered Attitude Control of a Rigid Spacecraft

Future spacecraft systems will require careful co-design over both physical and cyber elements to provide a better performance for holistic system. In this paper, a neural network (NN) based self-triggered control approach is introduced for the spacecraft attitude control problem. The self-triggered control is a resource-aware strategy which allows a reduction of the computation and communication demands, while still guaranteeing desirable closed-loop behavior. The NN is used for approximation of the triggered condition. We derive the equations of attitude motion for the spacecraft, and then develop a general procedure leading to NN based self-triggered feedback control implementations on a rigid spacecraft. Finally, the efficiency and feasibility of the obtained results are illustrated by means of a numerical spacecraft example.