Project Shadow Agent is an experimental project exploring new gameplay interactions using reinforcement learning. By working with Google Stadia, the team seeks new game genres and different potential applications.
Reinforcement learning (RL) is a relatively new machine learning method that learns the best actions based on reward or punishment in an environment. The goal of the project is to present a delightful user experience by creating believable characters in a game using existing machine learning tools to train different behaviors of Non-Player Characters (NPCs).