A Method of Attitude Control Based on Deep Deterministic Policy Gradient

The traditional methods of attitude control of satellite are represented by PID control, adaptive control, optimal control and intelligent control, etc. With these methods, lots of work of parameter adjustment and simulation needs to do on the earth. We proposed a method based on Deep Deterministic Policy Gradient (DDPG) to learn attitude control strategy in orbit in order to reduce the work and establish the ability of adapting to space environment. Through constructing training environment by using the attitude control system of satellite platform (ACSoSP), we trained an attitude control model and used the model to generate the strategy of attitude control. Validate the method by experiments in simulation environment.