Skip to content

s153625/5v5dota2ai-addon

 
 

Repository files navigation

The DotA 2 AI 5v5 Framework project

General disclaimer: this project is not affiliated with Valve. It's an academic research project.

About

A 5v5 framework developed by Kalle Lindqvist and Dennis Nilsson, an extension of a previous 1v1 framework developed by Tobias Mahlmann.

Goals of this project

To provide a framework able to to run 5v5 matches with bots in DotA 2. One team controlled by usermade bots competing against the in-game DotA 2 bots.

Submit the framework to competitions held by organizations such as the conference on games (COG) and foundations of digital games conference (FDG).

Using the framework

  Download DotA 2
  Download workshop tools for DotA2 (DLC)
  Download the 5v5 framework
  Copy the framework files located in the Dota2 AI Addon folder and put them in SteamLibrary/steamapps/common/dota 2 beta
  Run DotA 2 workshop tools, click play and select workshop tools
  Load the dota2ai project in the workshop tools
  Run the Python client, start it by running app.py in the Dota2 AI Framework folder (py app.py in a command prompt/terminal)
  Open vConsole in the workshop tool
  Run the following command in vConsole dota_launch_custom_game dota2ai dota to launch the mod

By performing these steps the framework will launch a game filled with bots of which the team good guys are controlled by example scripts that is included in the framework. The opposing team is controlled by in-game AI bots. You can find the example for controlling the bots in the Dota 2 AI framework/src folder. The example script currently shows a minimal bot implementation. Documentation regarding implementing more advanced features will be included. However by reading the code, a user should be able to implement a bot that can do anything possible in the game.

Future work

Implement functionality that supports 5v5 matches where all bots are controlled by user developed scripts.

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Packages

No packages published

Languages

  • Lua 99.6%
  • CSS 0.4%