Shadow Agent

Project Shadow Agent is an experimental game development project researching enhanced non-player character (NPC) behaviors using reinforcement learning tools. The goal of the project is to create a delightful experience consisting of more exciting interactions between NPCs and the environment. Reinforcement learning is a relatively new machine learning method which learns the best character behaviors based on rewards and punishments. By working with our client, Google Stadia, the team is exploring game genres and mechanics that prove reinforcement learning can be a useful tool in game design.

Team Members: Zoe (Zhiyu) Bai, Min Pan, Shitong Shen, Ava (ChenChen) Tan, Kristy (Yi Ting) Tsai

Project Instructors: Carl Rosendahl, Melanie Lam