Modern Robot Learning from Scratch - Lecture 2 - Robot Imitation Learning | Modern Robot Learning From Scratch

In this lecture, we will learn about the drawbacks of classical methods. Then we will understand the advantages of implicit methods over explicit methods. In the second phase of the lecture, we will learn about robot imitation learning. We will understand how Robot Imitation Learning works and how it is related to Supervised Learning. You will understand the meaning of observations and actions. Finally, we will understand how point estimate policies in robot imitation learning cannot deal with multi-modality. This will help us understand the need to design policies which can predict the probability distribution of actions instead of a single action for every observation.