A motivational model of action selection for virtual humans

Nowadays virtual humans such as non-player characters in computer games need to have a real autonomy in order to live their own life in persistent virtual worlds. When designing autonomous virtual humans, the action selection problem needs to be considered, as it is responsible for decision making at each moment in time. Action selection architectures for autonomous virtual humans should be individual, motivational, reactive and proactive to obtain a high degree of autonomy. This paper describes in detail our motivational model of action selection for autonomous virtual humans in which overlapping hierarchical classifier systems, working in parallel to generate coherent behavioral plans, are associated with the functionalities of a free flow hierarchy to give reactivity to the hierarchical system. Finally, results of our model in a complex simulated environment, with conflicting motivations, demonstrate that the model is sufficiently robust and flexible for designing motivational autonomous virtual humans in real-time

An action selection architecture for autonomous virtual humans in persistent worlds

Etienne De Sevin

Nowadays, virtual humans such as non-player characters in computer games need to have a strong autonomy in order to live their own life in persistent virtual worlds. When designing autonomous virtual humans, the action selection problem needs to be considered, as it is responsible for decision making at each moment in time. Indeed action selection architectures for autonomous virtual humans need to be reactive, proactive, motivational, and emotional to obtain a high degree of autonomy and individuality. The thesis can be divided into three parts. In the first part, we define each word of our title to precise their sense and raise the problematic of this work. We describe also inspirations from several domains that we used to design our model because this thesis is highly multi-disciplinary. Indeed, decision-making is essential for every autonomous entity and is studied in ethology, robotics, computer graphics, computer sciences, and cognitive sciences. However, we have chosen specific techniques to implement our model: hierarchical classifier systems and a free flow hierarchy. The second part of this thesis describes in detail our model of action selection for autonomous virtual humans. We use overlapping hierarchical classifier systems, working in parallel, to generate coherent behavioral plans. They are associated with the functionalities of a free flow hierarchy for the spreading of activation to give reactivity and flexibility to the hierarchical system. Moreover several functionalities are added to enhance and facilitate the choice of the most appropriate action at every time according to the internal and external influences. Finally, in the third part of this thesis, a complex simulated environment is created for testing the model and its functionalities with many conflicting motivations. Results demonstrate that the model is sufficiently efficient, robust and flexible for designing motivational autonomous virtual humans in persistent worlds. Moreover, we have just started to investigate on the emotional level which has to be improved in the future to have more subjective and adaptive behaviors and also manage social interactions with other virtual humans or users. Applied to video games, non player characters are more interesting and believable because they live their own life when people don't interact with them.

EPFL2006

An action selection architecture for autonomous virtual humans in persistent worlds

Etienne De Sevin

EPFL2006