Reinforcement Learning for Computational Guidance of Launch Vehicle Upper Stage

Table 2

Hyperparameters.

Parameter	Value	Parameter	Value
Shape reward coefficient	0.01	Epochs per update	30
Final reward coefficient	1000	Episodes per rollouts	50
Constant positive reward	0.001	Number of iterations	10000
Discount factor	0.995	Total episode	500000
GAE factor	0.98