In [1]:
# set up matplotlib and rendering
import matplotlib
import matplotlib.pyplot as plt
import matplotlib.animation as animation
from IPython import display
OpenAI Gym¶
https://gym.openai.com/docs/ simple tutorial
https://gym.openai.com/envs/#robotics Robotics environments
https://www.gymlibrary.ml/environments/mujoco/ Mujoco environments
https://www.gymlibrary.ml/ documentation
https://www.gymlibrary.ml/content/environment_creation/ Creating custom environments
In [2]:
import gym
display.HTML('<iframe width="560" height="315" src="https://www.youtube.com/embed/PGLzm-Gy0dQ" frameborder="0" allowfullscreen></iframe>')
/home/yasu/.local/lib/python3.10/site-packages/gym/wrappers/monitoring/video_recorder.py:9: DeprecationWarning: The distutils package is deprecated and slated for removal in Python 3.12. Use setuptools or check PEP 632 for potential alternatives
import distutils.spawn
/home/yasu/.local/lib/python3.10/site-packages/IPython/core/display.py:419: UserWarning: Consider using IPython.display.IFrame instead
warnings.warn("Consider using IPython.display.IFrame instead")
Out[2]:
MuJoCo environment¶
mujoco environment for gym is probably already prepared if you have mujoco-py
But for some weird reason, if the other environments are run before this, notebook kernel fails...
In [3]:
env = gym.make('Ant-v2')
env.reset()
fig = plt.figure()
ims = []
for _ in range(20):
ims.append([plt.imshow(env.render(mode='rgb_array', width=256, height=256))])
env.step(env.action_space.sample()) # take a random action
env.close()
ani = animation.ArtistAnimation(fig, ims, interval=50)
html = display.HTML(ani.to_jshtml())
display.display(html)
plt.close()
/home/yasu/.local/lib/python3.10/site-packages/gym/envs/registration.py:505: UserWarning: WARN: The environment Ant-v2 is out of date. You should consider upgrading to version `v3` with the environment ID `Ant-v3`.
logger.warn(
/home/yasu/.local/lib/python3.10/site-packages/mujoco_py/builder.py:9: DeprecationWarning: The distutils.sysconfig module is deprecated, use sysconfig instead
from distutils.sysconfig import customize_compiler
<frozen importlib._bootstrap>:283: DeprecationWarning: the load_module() method is deprecated and slated for removal in Python 3.12; use exec_module() instead
Robotics environment¶
pip3 install gym-robotics
This must also be run before the other environments...
In [4]:
env = gym.make('HandManipulateBlock-v0')
env.reset()
fig = plt.figure()
ims = []
for _ in range(20):
ims.append([plt.imshow(env.render(mode='rgb_array', width=256, height=256))])
env.step(env.action_space.sample()) # take a random action
env.close()
ani = animation.ArtistAnimation(fig, ims, interval=50)
html = display.HTML(ani.to_jshtml())
display.display(html)
plt.close()
Box2D environments¶
https://www.gymlibrary.ml/environments/box2d/
to get the Box2D environment...
sudo apt install swig
pip3 install box2d
pip3 install box2d-py
In [5]:
env = gym.make('CarRacing-v1')
env.reset()
fig = plt.figure()
ims = []
for _ in range(20):
ims.append([plt.imshow(env.render(mode='rgb_array'))])
env.step(env.action_space.sample()) # take a random action
env.close()
ani = animation.ArtistAnimation(fig, ims, interval=50)
html = display.HTML(ani.to_jshtml())
display.display(html)
plt.close()
Track generation: 1021..1288 -> 267-tiles track
In [6]:
env = gym.make('BipedalWalker-v3')
env.reset()
fig = plt.figure()
ims = []
for _ in range(20):
ims.append([plt.imshow(env.render(mode='rgb_array'))])
env.step(env.action_space.sample()) # take a random action
env.close()
ani = animation.ArtistAnimation(fig, ims, interval=50)
html = display.HTML(ani.to_jshtml())
display.display(html)
plt.close()
In [7]:
env = gym.make('CartPole-v1')
env.reset()
fig = plt.figure()
ims = []
for _ in range(50):
ims.append([plt.imshow(env.render(mode='rgb_array'))])
env.step(env.action_space.sample()) # take a random action
env.close()
ani = animation.ArtistAnimation(fig, ims, interval=50)
html = display.HTML(ani.to_jshtml())
display.display(html)
plt.close()
/home/yasu/.local/lib/python3.10/site-packages/gym/envs/classic_control/cartpole.py:163: UserWarning: WARN: You are calling 'step()' even though this environment has already returned done = True. You should always call 'reset()' once you receive 'done = True' -- any further steps are undefined behavior.
logger.warn(
In [8]:
env = gym.make('MountainCar-v0')
env.reset()
fig = plt.figure()
ims = []
for _ in range(50):
ims.append([plt.imshow(env.render(mode='rgb_array'))])
env.step(env.action_space.sample()) # take a random action
env.close()
ani = animation.ArtistAnimation(fig, ims, interval=50)
html = display.HTML(ani.to_jshtml())
display.display(html)
plt.close()
let's look at the action / observation space, and what env.step returns...
In [9]:
env = gym.make('CartPole-v1')
env.reset()
observation, info = env.reset(return_info=True)
print(f"info:{info}")
print(f"action space:{env.action_space}")
# one action pushes cart to right and the other to the left
print(f"observation space:{env.observation_space}")
print(f"reward range:{env.reward_range}")
for i in range(3):
observation, reward, done, info = env.step(env.action_space.sample()) # take a random action
print("-"*10)
print(f"observation:{observation}")
print(f"reward:{reward}")
print(f"done:{done}")
print(f"info:{info}")
env.close()
info:{}
action space:Discrete(2)
observation space:Box([-4.8000002e+00 -3.4028235e+38 -4.1887903e-01 -3.4028235e+38], [4.8000002e+00 3.4028235e+38 4.1887903e-01 3.4028235e+38], (4,), float32)
reward range:(-inf, inf)
----------
observation:[-0.00195471 -0.17760813 0.0406442 0.35323688]
reward:1.0
done:False
info:{}
----------
observation:[-0.00550687 -0.37328374 0.04770894 0.65845406]
reward:1.0
done:False
info:{}
----------
observation:[-0.01297255 -0.5690361 0.06087802 0.9657696 ]
reward:1.0
done:False
info:{}
In [10]:
from gym import spaces
space = spaces.Discrete(8)
print(space)
print(space.contains(2))
print(space.contains(8))
print(space.n)
Discrete(8) True False 8
Control environment with keyboard input
In [11]:
import pygame
from gym.utils.play import play
mapping = {(pygame.K_UP,): 2, (pygame.K_DOWN,): 3}
env = gym.make("Pong-ram-v0")
print(env.unwrapped.get_action_meanings())
play(env, keys_to_action=mapping)
['NOOP', 'FIRE', 'RIGHT', 'LEFT', 'RIGHTFIRE', 'LEFTFIRE']
/home/yasu/.local/lib/python3.10/site-packages/gym/utils/play.py:11: UserWarning: WARN: failed to set matplotlib backend, plotting will not work: No module named 'tkinter' logger.warn(f"failed to set matplotlib backend, plotting will not work: {str(e)}") /home/yasu/.local/lib/python3.10/site-packages/gym/envs/registration.py:505: UserWarning: WARN: The environment Pong-ram-v0 is out of date. You should consider upgrading to version `v5` with the environment ID `ALE/Pong-ram-v5`. logger.warn( A.L.E: Arcade Learning Environment (version 0.7.5+db37282) [Powered by Stella]