Reentry Trajectory Planning Based on Proximal Policy Optimization

In order to deal with the changing flight environment in the process of hypersonic vehicle reentry, a real-time trajectory planning algorithm based on Proximal Policy Optimization (PPO) was proposed. The proximal policy optimization is used to train the hypersonic vehicle reentry process, and the intelligent vehicle that can obtain the best trajectory control output according to the state input is obtained. Continuous roll Angle, discrete roll Angle, and continuous roll Angle change rate are selected as actions to study their training effects on the reentry process. The results show that the action space based on the change rate of the roll Angle converges faster and the reentry flight time is shorter. The trajectory planning method based on proximal strategy optimization can quickly generate the optimal trajectory of high speed aircraft. Compared with the trajectory planning algorithm based on pseudospectral method, the proposed method has the generalization ability to meet the accuracy requirements and can meet the needs of online real-time trajectory planning.