Hands-On Q-Learning with Python

Creating a task loop

In this section, we'll create our first task loop. We'll let our agent run by making random moves until it successfully reaches the goal of dropping off a passenger at the correct location.

Here is a simple way to implement a randomly-acting agent:

state = env.reset()reward = 0while reward != 20:    observation, reward, done, info = env.step(env.action_space.sample())env.render()

We reset the environment to put it into a random state for the start of the loop. We set the reward to zero, meaning that the goal has not yet been reached. We call env.step() repeatedly, with env.action_space.sample() as an argument to get a random action from the six allowable actions, and then advance the agent by one step. Finally, we set the ...