Improving Computational Efficiency for Powered Descent Guidance via Transformer-based Tight Constraint Prediction