Skip to content

Agent-76 is a Reinforcement Learning based A.I. that learns to play Overwatch 2 by itself.

License

Notifications You must be signed in to change notification settings

AgamChopra/Agent-76

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

18 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Agent-76: Reinforcement Learning-Based Autonomous Overwatch 2 Player

Agent-76 is an innovative Reinforcement Learning-driven Artificial Intelligence designed to autonomously master the intricacies of Overwatch 2, a highly popular First Person Shooter game. Overwatch 2 features a dynamic gaming environment with a complex state and action space, necessitating the development of a robust policy to adapt swiftly to the ever-evolving in-game situations.

Named after the iconic character "Soldier 76" from the game, Agent-76's primary objective is to maximize its in-game rewards while learning to navigate the complex environment and engage with other players effectively.

Note: In line with ethical considerations, this bot exclusively participates in games with in-game AI and human players in custom competitive lobbies, with their consent. This collaboration enhances the project's development and maintains a fair gaming environment.

I emphasize that this project is entirely open-source. I kindly request that you refrain from misusing the system. Utilizing this bot to play on your account in public lobbies constitutes cheating and could result in a breach of terms and conditions, potentially leading to a permanent account ban. Please understand that I will not make the trained parameters publicly available for the same reasons.

Agent-76 is intended primarily as an educational endeavor, promoting the exploration of Reinforcement Learning and Artificial Intelligence concepts within the gaming domain.

About

Agent-76 is a Reinforcement Learning based A.I. that learns to play Overwatch 2 by itself.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published