Baby Hippo Spring - Design Documentation
Exploring reinforcement learning agents in game experience
Introduction
Baby Hippo Spring is an experimental game exploring the application of reinforcement learning (RL) in non-player characters (NPC). Challenged with the uncertainty and unpredictability of results from RL training, we aim to make use of this essence of RL to build a delightful game experience for players.
In this documentation, you will learn how we cracked the challenges of reinforcement learning in games, where our inspirations came from, our design process, experiments, findings, and lessons learned.
About us
We are Team Shadow Agent, a multi disciplinary team of 5 second-year graduate students from Entertainment Technology Center, Carnegie Mellon University - Silicon Valley Campus. Our client is Erin Hoffman-John and her team from Google Stadia. Our faculty advisors are Carl Rosendahl and Melanie Lam.
