Description
Reinforcement Learning (RL) is a method of training an AI agent using a system of rewards to complete a task or goal. Google used Reinforcement Learning in its AlphaGO project and there are countless examples of RL agents learning to play various classic Atari games with a recent RL agent reaching grandmaster level in StarCraft II.
The Unity game engine is very popular and has a freely available ML-Agents Toolkit for experimenting with AI. Unity is a popular choice for this type of research as it has a physics engine built-in. This allows training of an AI in a simulated environment for hours or days before applying to a robot or other autonomous system in the real-world.
In this hands-on tutorial, we’ll build a simple environment using the Unity game engine and a basic RL agent that attempts to reach a goal. This is a follow-along style and no coding experience is needed. This is a great workshop for educators as a way to get their students excited about AI research.
Prerequisites
Attendees should have a laptop with Unity installed as well as Python (which are both free to install) before the event. Instructional video for installing Unity: http://bit.ly/before_we_start
Organizers
- Carmine T. Guida, Pace University