Development of a web based tool intended to animate secondary (or background) 3D virtual agents (where quantity is more important than quality) in real time, by giving them a behaviour, through node-based programming, that makes them aware of their surroundings and can navigate in a scenery and interact with points of interesnt and other virtual agents semi-autonomously. Also the goal is to achieve behaviors that differs from the rest of the entities even though the action is the same, giving some personality through an emotional body pose based stylization of the animations, taking into account character properties, and finally having a realistic background scene, populated with each person behaving uniquely.
The advantage of this tool is that the artists do not have to worry about the individual behavior of each secondary agent, as it is automatically managed by our AI generated by an hybrid approach of Behavior Trees. Also, this behaviors will become reusable once they are created, as other artists may want to generate similar scenarios or they do not want to start from zero and want to have a base behavior to continue developing theirs, exportable to external systems, to embedd the core library (HBTree.js) in their pwn projects and use the Hybrid Behaviour Trees.
The application can be tested on this link. It is continuously growing and changing.
Also it is not perfectly improved, so the performance can be damaged if the number of agents increases dramatically.
This project has been developed within the context of the SAUCE Project, a Horizon 20202 EU project.