import gym # pip install -U gym pygame

# https://gym.openai.com/docs/

env = gym.make('CartPole-v1')

env.action_space # Discrete(2)

obs = env.reset() # [-0.01258566, -0.00156614, 0.04207708, -0.00180545]

env.render()

img = env.render(mode='rgb_array')

img.shape # (800, 1200, 3) - height, width, channels (3 = Red, Green, Blue)

action = 1 # accelerate right

obs, reward, done, info = env.step(action)

obs # [-0.01261699, 0.19292789, 0.04204097, -0.28092127]

reward # 1.0

done # False

info #{}

# basic example

env = gym.make('CartPole-v1')

observation = env.reset()

for _ in range(1000):

env.render()

action = env.action_space.sample() # your agent here (this takes random actions)

observation, reward, done, info = env.step(action)

if done:

observation = env.reset()

env.close()