Researchers from the STAMINA4Space Program and PhilSA investigated how Deep Reinforcement Learning (RL) could be used for the attitude control of satellites. In their work, two RL algorithms were implemented: Proximal Policy Optimization (PPO) and Soft Actor-Critic (SAC). The algorithms were trained in a simulated environment using the Diwata microsatellite simulator, MATA: Mission, Attitude, and Telemetry Analysis software. The performance of the algorithms were compared with that of Diwata’s Proportional-Integral-Derivative (PID) control. The results obtained by the group show that RL can outperform traditional controllers in terms of settling time, overshoot, and stability. The results of this research will help solve problems in conventional attitude controllers and enable satellite engineers to design a better Attitude Determination and Control System (ADCS).
The paper was presented at the 35th Small Satellite Conference.