ADAPTIVE CONTROL FOR PROXIMITY MANEUVERS AROUND ASTEROIDS USING EXTREME LEARNING MACHINE-BASED GRAVITY FIELD ESTIMATION

Proximity operations around asteroids are challenging because of the difficulties in the real-time estimation of their gravity field. Additionally, it is extremely important to design control strategies to perform essential maneuvers. In this paper, direct adaptive control is employed to study proximity maneuvers around asteroids, such as hovering and orbit station keeping. Thanks to this approach, the control law is directly adjusted to minimize the error between plant and reference trajectory outputs. The control law’s gains are tuned to obtain the desired performances. Furthermore, to consider a fast and accurate estimation of the gravity, Extreme Learning Machine is exploited to map the position vector into the gravity vector in the asteroid body-fixed frame. Finally, all the framework is applied to proximity maneuvers around asteroid 951 Gaspra, obtaining good results.