Python Reinforcement Learning Projects

The classic control tasks

We will now go through the details of each control task and answer the following questions:

1. What are the control inputs and the corresponding feedbacks?
2. How is the reward function defined?
3. Is the action space continuous or discrete?

Understanding the details of these control tasks is quite important for designing proper reinforcement learning algorithms because their specifications, such as the dimension of the action space and the reward function, can affect the performance a lot.

CartPole is quite a famous control task in both the control and reinforcement learning communities. Gym implements the CartPole system described by Barto, Sutton, and Anderson in their paper Neuronlike Adaptive Elements That Can Solve ...