logo logo
Home arrow Conscious Machines... arrow Machine Consciousness Projects arrow AI Bots in Video Games and Virtual Worlds
Sunday, 29 November 2015
AI Bots in Video Games and Virtual Worlds Print E-mail
Written by Raúl Arrabales Moreno   
Wednesday, 10 December 2008

AI Bots in Video Games and Virtual Worlds

Usually when we talk about situated agents as the target of the research in Artificial Intelligence or Machine Consciousness, we think about physical agents, like typical autonomous robots. One of the reasons why we tend to use physical robots as part of our experimental setup is because we believe embodiment plays a key role both in intelligence and consciousness.

AI Bots in a video game
AI Bots in a video game

Because of the limitations in cost and time, during development phases we are used to using simulation tools in order to quickly test our hypotheses. However, the final target is always the physical robot and its application into the real world. At least that is the obvious conclusion in the field of robotics.

But, what about final AI applications that only live within virtual worlds? Do they deserve less attention from AI research fields? In recent years we are seeing a growing interest in applications which reside entirely within virtual worlds and video games. Some relevant examples are Second Life and World of Warcraft. I personally believe the success of these products is rooted in the fact that they provide new ways of interaction between humans (players, colleagues, partners, etc.). We currently lack the same level and richness of interaction when it comes to AI bots (we use the term AI bot to refer to autonomous virtual agents that are controlled by an AI program).

From my point of view, there is no doubt that AI bots are a new example of situated agents. Whether they can be considered embodied or not is another question. Anyhow, we could say that they are embodied in terms of the simulated physical laws enforced by the engine which generates the corresponding virtual world or video game.

Development resources for AI research with AI Bots

As I’ve been recently looking for a rich simulation environment for testing my Machine Consciousness architecture as the control of an AI Bot, I put together a short list of engines, SDK, and development tools (take into account that I am not considering here any robotics development framework like MRDS or URBI, also I am excluding tools for complete game development. I just want to focus specifically on AI Bot programming).

Video Games

In the video games arena, in principle any liberated game (source code available) could be suitable for experimentation with AI autonomous players. A classical examples would be the famous Quake (Quake C source code is available as GPL from ID Software). For a list of liberated games visit http://liberatedgames.com/

Having the source code is not a requirement in order to program AI Bots for a game. Actually, if we want to develop AI bots for most current video games, it is very likely that we don’t have access to the source code. However, most of current video games provide tools for adding game features like maps, bots, mods, etc. Some of these tools are particularly useful to develop external interfaces, so we can run the AI Bots logic externally. This is also particularly useful if we want to use a different programming language for our AI bots.

A good example of this flexibility is the Unreal Tournament series by Epic Games. The game comes with a programming language called UT Script that has allowed developers to write mutators. In the case of Unreal Tournament, the GameBots mutator permits to retrieve game state and send commands to the game server.

The GameBots component has been used by some other developers to build new API extensions so you can develop your AI Bot using Java or C#. Have a look to Pogamut 2 for Java programming of AI Bots in UT 2004 and    
UT3 .NET Bots for .NET programming for Unreal Tournament 3.

ORTS is another sort of example. In this case, a Real-Time Strategy game. While First-Person Shooter (FPS) video games are suitable for some sort of AI research, the RTS environments are also a good tool for multi-agent systems and planning problems.

Virtual Worlds

Actually, video games and virtual worlds are quite related when we talk about MMOG (Massively Multiplayer Online Games). I think Second Life is the most known example. If you want to experiment and develop applications in this framework you can check libsecondlife.

Multiverese is also a product to check. It is a platform for the development of virtual worlds and MMOGs (it is based on the OGRE open source graphics rendering engine).

realXtend is another open source platform for the creation of virtual worlds. It is based on OpenSim (an open source 3D application server).


Add as favourites (614) | Quote this article on your site | Views: 10599

  Be first to comment this article
RSS comments

Only registered users can write comments.
Please login or register.

Related Items:

  1. MCexperts
  2. Machine Consciousness Bibliography
  3. Consciousness in Vegetative State?
  4. The Quest for Consciousness: A Neurobiological Approach
  5. Consciousness and the Biology of the Brain
  6. Experiments in Consciousness
  7. Elephants recognize themselves in the mirror
  8. Raúl Arrabales Moreno
  9. Igor Aleksander
  10. Owen Holland
Last Updated ( Wednesday, 10 December 2008 )

Lost Password?
 Conscious Robots RSS FeedConscious Robots RSS Feed

Find us on Facebook

Follow us on TwitterFollow us on twitter

Machine Consciousness Bibliography Database


The Cognitive Machine Consciousness Scale

Last Posts in Forum
miel continental