Exjobbsförslag från företag

Detta är ett uppsatsförslag hämtat från Nationella Exjobb-poolen. Klicka här för att komma tillbaka till samtliga exjobbsförslag.

Förslaget inkom 2009-02-18

Cognitive Models for Virtual Characters

OBS! ANSÖKNINGSTIDEN FÖR DETTA EXJOBB HAR LÖPT UT.
The rapid development of processing and rendering speeds on commodity desktop computers is enabling the development of very realistic real-time computer generated environments. In the past few years we have seen the development from realistic looking static environments towards increasingly dynamic environments. The first target has been physical simulation of the effects such as moving, deforming and even breaking objects. Then water, fire and even hair and clothing on simple virtual characters are becoming increasingly realistic. A challenge in the quest for realism now lies in equipping the virtual characters populating virtual worlds with a higher level of intelligence.

Aim
The focus of this project is to deploy state-of-the-art artificial intelligence methods into character simulation and extract and address new researchers issues in artificial intelligence in the context of the developed applications.We consider virtual, autonomous characters situated in dynamic, unpredictable, virtual worlds. To render these characters self-animated (i.e., alive) in real time performances, we propose a new approach to make them able to perceive, reason and act in the world where they are situated. This goal can be achieved by ascribing characters cognitive capabilities together with reactive functionalities that implement primitive behaviors (such as avoiding collisions). The specific aims of the project are to develop:

- cognitive models for virtual characters;

- a model for perception and actuation, that is a model to couple the world and the mental state of the characters;

- an architecture (for characters) combining high-level cognitive capabilities with low-level reactive functionalities (such architecture should be parameterizable, and hence flexible and adjustable);

- techniques for setting up (updatable) character groups and societies, and for combining groups into larger ones;

- techniques to validate both characters behavioral properties and global properties of virtual societies;

- an implementation of the cognitive capabilities and the character's architecture.


State-of-the-art
Computer animation is concerned with producing sequences of images that when displayed at high speed give the illusion that certain components of the image move. The aim is to produce animations that look realistic. Computer animation has typically focused on low-level locomotion problems, i.e., low-level control.

In the mid-1980s, researchers began incorporating physical principles to develop physical models for animating passive objects, such as falling and colliding objects. The idea behind this kind of model was to explicitly represent physical concepts. In the last decades, researchers started to focus on the requirement that characters in animations should behave realistically. That is, characters should be able to perform sequences of movements. This is commonly referred to as high-level control problem. Research in behavioral modelling progressed toward self-animating characters that react appropriately to stimuli perceived from the environment.

At the top of the modelling pyramid, cognitive modelling has emerged as the use of artificial intelligence techniques, including knowledge representation, reasoning, and planning, to produce virtual characters with some level of deliberative intelligence. Addressing human cognitive functionality is a challenging research area in artificial intelligence. Funge was pioneering the use of hardcore artificial intelligence techniques in computer games and animation. His aim was to devise a system suitable for rapid prototyping and producing off-line animations. To do so, he used logical reasoning to shift most of the work for generating behavior from the animator to the animated characters. With the intention that the devised system should be easy to build, reconfigure and extend.

Collaborations
* Prof. Luis Moniz Pereira, Director of Center of Artificial Intelligence C

  GÅ TILL XJOBB.NU FÖR FULLSTÄNDIG INFO OM DETTA EXJOBB




Informationen om uppsatsförslag är hämtad från Nationella Exjobb-poolen.