1. Learn
    2. /
  2. Courses
    3. /
  3. Reinforcement Learning with Gymnasium in Python
Connected

Exercise

Improving a policy

In the previous exercise, you computed the Q-values for each state-action pair in the MyGridWorld environment. Now, you'll use these Q-values to improve the existing policy. Policy improvement is a critical step in reinforcement learning, where you enhance the policy by choosing actions that maximize the expected utility (Q-value) in each state. After improving the policy, you will render the new movements according to this improved policy.

The environment has been imported as env, along with the Q-values as Q, and the render() function.

Instructions

100 XP
  • Find the best action for each state based on Q-values.
  • Select the right action based on the improved_policy.
  • Execute the selected action to observe its outcome.