dm_control_cartpole_environment_random_action_loop.py

1from dm_control import suite
2import numpy as np
3
4# Load one of our environments:
5env = suite.load(domain_name="cartpole", task_name="swingup")
6
7# Iterate over a task episode:
8time_step = env.reset()
9while not time_step.last():
10  action_spec = env.action_spec()
11  action = np.random.uniform(action_spec.minimum,
12                             action_spec.maximum,
13                             size=action_spec.shape)
14  time_step = env.step(action)
15  print(time_step.reward, time_step.discount, time_step.observation)