gym-ao
is a simulation framework based on Gymnasium for testing Reinforcement Learning (RL) algorithms for Adaptive Optics (AO), specifically for focal plane wavefront control.
Two environments have been developed:
gym_sharpening.py
: Focal plane wavefront control with the goal of maximizing the Strehl ratio based on focal plane images.gym_darkhole.py
: Dark hole digging based on pairwise probing with the goal of maximizing contrast.
from gym_ao.gym_sharpening import *
def run_sharpening():
env = Sharpening_AO_system()
N_iter = 100
N_episode = 10
for episode in range(N_episode):
o = env.reset()
print('Episode:', env.episode)
for i in range(N_iter):
a = 0.1 * env.action_space.sample()
o, r, t, trunc, info = env.step(a)
if trunc:
break
env.render()
env.close()
if __name__ == '__main__':
run_sharpening()