Learning prediction-correction guidance for impact time control

This paper investigates the problem of impact-time-control and proposes a learning-based computational guidance algorithm to solve this problem. The proposed is developed based on general prediction-correction concept: exact time-to-go under proportional navigation with realistic aerodynamic characteristics estimated by deep neural network biased command nullify impact time error utilizing emerging reinforcement learning techniques. To deal insufficient training data, transfer-ensemble approach train network. augmented into block resolve issue sparse reward that has been observed in typical formulation. Extensive numerical simulations are conducted support algorithm.