Usually, people often confuse reinforcement learning with supervised learning. Reinforcement learning is a learning technique where the machine learns on the go and doesn't need training data, unlike supervised learning. The machine tries to understand the application by interacting with the application directly and stores a history of information, along with a reward system that indicates whether the machine did the right thing or not.

Reinforcement learning can be found in games, where the computer tries to understand the user's action and reacts accordingly to ensure that it wins. Reinforcement learning is also being tested out in self-driving cars and robotics. As this chapter specifically deals with robotics, we shall look into this learning technique more deeply. Reinforcement learning can become supervised learning if the machine has understood enough from the environment it is interacting with. You shall understand this more clearly in the upcoming sections when we talk about explore versus exploit.

Now, let's learn more about reinforcement learning, which will be the main concept of this chapter.