Teaching Video Game Characters Natural Body Language

Video game characters with natural responses to human body language

Researchers at Goldsmiths, University of London have been using theater performers to design computer software capable of reading and replicating the way in which humans communicate with their bodies.

Dr Marco Gillies from the Department of Computing has made virtual characters more believable by enlisting actors to teach them body movement. The actors interact with members of the public through a screen, and their responses to specific body language are memorized as algorithms by the software.

“Two people can take on the roles of the video game character and the player, showing how the character should respond by acting out the movements themselves,” explained Dr Gillies. “The software enables video games characters to move in a more natural way, responding to the player’s own body language rather than mathematical rules.”

Traditionally, the creators of interactive characters are computer programmers, but Dr Gillies and his team puts this task in the hands of people with artistic rather than technical knowledge.

“Our hypothesis is that the actors’ artistic understanding of human behavior will bring an individuality, subtlety and nuance to the character that it would be difficult to create in hand authored models,” said Dr Gillies. “These are the kinds of everyday movements, that we do unconsciously, which make them hard to program in the conventional way.”

Dr Gillies and his team set up a case study in which physical theater performer Emanuele Nargi taught the software natural responses to a player’s movement.

One of the players, Goldsmiths student Max Bye, noticed that the virtual character reacted in a human manner: “When I laughed at it, it would walk away disappointed, so that worked very well.”

The research intends to help interactive media represent more nuanced social interaction, broadening its range of application. The new technique optimizes the use of the latest generation of motion detectors, and it is hoped that in future this will lead to games that are more emotionally complex and able to respond to more subtle social nuances of human behavior.

Written by: Goldsmiths, University of London (via Presence)
Posted by: Situated Research

This Post Has 7 Comments

  1. XFunc_CaRteR

    My God… Why don’t they just use actors?

    No computer can “can” the creativity that a real human being can deliver. Why? Because it is always based on data – which is to say, it is based on the past. And creativity is about bringing the new out of thin air.

    Just use an actor.

    1. Matthew Sharritt, Ph.D.

      Thanks for your comment. You make a good point, however all future games and character engagement would then have to be ‘acted’ and (pre)scripted. By creating models of common human emotions, the game might achieve a more adaptable storyline, as well as less acting needed for future games since complex non-verbal human communication models would be at the design team’s disposal.

  2. XFunc_CaRteR

    Less acting needed?

    Why would I take the human body language creativity away from an expert in it – an actor – and give it to a non-expert – a programmer or a game designer?

    That’s like developing a program to allow a non-musician to pre-can music, so you can get rid of creative musicians.

    It makes no sense, whatsoever.

    1. Matthew Sharritt, Ph.D.

      Acting requires an understanding of complex human communication. However, in the case of movies, for instance, they are meant for the masses. The masses are able to digest it because we are all social, and participate in communication every day with others. Game designers are no exception. If game designers didn’t understand anything about communication it’d be impossible to make engaging games.

      I think the point is to try and model human communication and impart some of those properties to game characters, such that more ‘acting’ can be done in a dynamic fashion, rather than just game cut-scenes. Games differ from movies in that they are dynamic, branching narratives, and a prescribed (acting) script cannot always suffice. The more dynamic the narrative, the harder to script and act beforehand.

  3. XFunc_CaRteR

    I might add that doing *anything* that “cans creativity” is to immediately kill creativity.

    Creativity *depends* on reinvention. It cannot survive solely on patterns that were created (and subsequently canned, if you will) in the past. It thrives on the novel and the new. Only human beings can do this. And, in the case of actings, only actors can do it.

    1. Matthew Sharritt, Ph.D.

      Agreed, canning creativity can be disastrous. In order to create game avatars that continually improve (and make the game better), some sort of AI / machine learning might help. That was the point of the article, I believe. If games can learn the subtleties of human communication, game characters can respond to gamers in more realistic and engaging ways by emulating that nuanced behavior.

  4. XFunc_CaRteR

    AI cannot learn the subtley of human communication because… wait for it… It *ain’t human*!

    It’s only human communication when a human does the communicating! Right!?!

Leave a Reply