This project allows the creation of agents and their behavior with the objective of creating more believable interactions.

This project is based on a Master's thesis that defends that the subdivision of an action into several stages increases the immersion of the audience in a scene.

Our agent’s behavior is a 4-step anticipation-based cycle: (1) the agent perceives changes in the world based on its expectations, (2) reacts to them emotionally based on what is anticipated, (3) decides what to do next and anticipates what will happen, (4) and performs part of an on-going action, then repeats. This cycle occurs multiple times in the course of an action allowing the virtual agents to perceive events and express different emotions, among other things, while performing a same action that gives context to that expression.

3Motion differentiates itself from standard approaches in how it de-constructs the traditional atomic action generally used when implementing virtual agents behavior,into three explicit and distinct stages: anticipation, action, and follow-through. Each stage may take a certain time to play out, and interaction at different points in the sequence will have a different meaning for the other agents participating in the scene, as well as for the viewer passively watching or actively interacting with the scene, allowing for the creation of a richer interaction.

This work was partially supported by national funds through Fundacão para a Ciência e a Tecnologia (FCT) with ref. UID/CEC/50021/2013, and FCT grant from project Tutoria Virtual with ref. TDC/IVC-PEC/3963/2014.