Diving into Reinforcement Learning can seem daunting if you do not have the proper hands on guidance. Many times, people have asked me whether they should master Deep Learning before delving into Reinforcement Learning and my answer has always been that “it depends on what you want to do with RL”. RL is a broad domain in its own respect. There are classical RL algorithms that can be learned and applied without any Deep Learning experience. There is also Deep Reinforcement Learning which leverages neural networks to help RL agents to learn proper behaviors in their environment through trials and error with reward functions.

This course has been designed to be the easiest and fastest basic entry point into RL and its applications in Robotics. From the first to the last videos, I explain every concept in RL in the context of robotics. I am intentional about this because I want to empower you to readily know when and how to apply RL techniques in Robotics. It is not an advanced course. Instead, it is your best option when you are getting started in RL (without any prior knowledge) and you are interested in being able to readily apply what you learn in your robotic projects.

Even though I use sensors and actuators from the EV3 Mindstorms robotics kit in the hands on implementation sessions, you do not necessarily have to purpose the kit to get the best out of this course. It will certainly enhance your learning experience if you have the kit but do not worry if you do not. Without the kit, you can still understand the concepts and apply them on any robotic platform. It is my desire that after finishing this course, your passion for RL will be ignited and you will study further to know about more advanced algorithms and techniques.

So, while this course would not teach you how to build superhuman capabilities into your robotics projects, you will certainly learn how to program robots to behave well in their environments without explicit instructions on what is considered “proper behavior”.
–
SKILLS YOU WILL GAIN
Using Python to control sensor and actuators on EV3 Mindstorms robots.
Developing reward functions for Reinforcement Learning in robotic contexts.
Applying the Multi-Armed Bandit algorithm to help robots to automatically learn expected behaviours.
WHAT YOU WILL LEARN
Balancing exploration and exploitation using various strategies.
Understanding the multi-armed bandit algorithm.
Programming simple robots that can automatically learn the best actions to take in their environments.