Training a cheetah bot to run using PPO

In this section, let's learn how to train the 2D cheetah bot to run using Proximal Policy Optimization (PPO). First, import the necessary libraries:

import gym
from stable_baselines.common.policies import MlpPolicy
from stable_baselines.common.vec_env import DummyVecEnv, VecNormalize
from stable_baselines import PPO2

Create a vectorized environment using DummyVecEnv:

env = DummyVecEnv([lambda: gym.make("HalfCheetah-v2")])

Normalize the state:

env = VecNormalize(env,norm_obs=True)

Instantiate the agent:

agent = PPO2(MlpPolicy, env)

Train the agent:

agent.learn(total_timesteps=250000)

After training, we can see how our trained cheetah bot learned to run by rendering the environment:

state = env.reset()
while True:
    action, _ = agent.predict(state)
    next_state, reward, done, info = env.step(action)
    state = next_state
    env.render()

Save the whole code used in this...