dm_control_suite_cartpole_random_action_simulation_loop.py

1from dm_control import suite
2import numpy as np
3
4# Load one of the standard environments
5env = suite.load(domain_name="cartpole", task_name="swingup")
6
7# Iterate over a few steps
8action_spec = env.action_spec()
9time_step = env.reset()
10
11while not time_step.last():
12    # Generate a random action
13    action = np.random.uniform(action_spec.minimum,
14                               action_spec.maximum,
15                               size=action_spec.shape)
16    
17    # Step the environment
18    time_step = env.step(action)
19    
20    # Print the reward and discount
21    print(f"Reward: {time_step.reward}, Discount: {time_step.discount}")