PRESENT moves forward with the creation of the Behavior Planner to develop agents with verbal and nonverbal behaviours

A research and innovation project in artificial intelligence belonging to the European Commission’s H2020 Programme, coordinated by Josep Blat and Vanesa Daza. The GTI has developed a web-based editor that allows creating an agent with verbal and nonverbal behaviours.

Josep Blat and Vanesa Daza , director of the Interactive Technologies research group ( GTI ) and researcher at the Wireless & Secure Communications Research Group (WiSeCom) ( WiCom ), respectively, at the UPF Department of Information and Communication Technologies ( DTIC ), are coordinating Photoreal REaltime Sentient ENTity ( PRESENT ), selected in the H2020/ICT/2018/3 call and is to run for three years.

PRESENT is currently half-completed and has achieved promising results. The Interactive Technologies research group (GTI) led by Josep Blat attended the review of the PRESENT project on 28 April, a year and a half after its inception.

Together with other project partners, the GTI presented its work on PRESENT to the reviewers and the project manager. The progress achieved was very highly appraised, as "the results are very promising and there is clear teamwork among partners that will surely result in a successful project", they asserted in the course of the project meeting.

Behavior Planner, a web-based editor that allows creating an agent capable of complex behaviours that produce both verbal and nonverbal reactions

"At the GTI, our approach focuses on human-machine interaction and 3D graphics on the web. The GTI has presented our progress with the Behavior Planner, a web-based editor that allows creating an agent capable of complex behaviours that produces both verbal and nonverbal reactions, that is, apart from specifying what to say, it also describes the gestures and animations it must make", the GTI researchers state. This application has been made possible thanks to Medusa, a node-based system to produce behaviour trees created by the GTI in the SAUCE project", they added.

Thus, in the PRESENT project, the GTI makes two main contributions: the first is to control the verbal and nonverbal behaviour of the virtual agents. To do so, as mentioned above, they have developed a web-based editor that allows creating behaviour trees using node-based programming; secondly, it enables developing low-end rendering on the web using advanced techniques of lighting and post-processing effects.

Developing virtual digital companions

The main goal of PRESENT is to develop virtual digital companions -sentient agents- provided with a high degree of realism and naturalness and emotional sensitivity that will be able to engage in meaningful dialogues, provide meaning to the experience and act highly reliably as tutors and guides for Augmented and Virtual Reality interfaces, among others.

The realistic virtual agents resulting from PRESENT may be used for a wide range of applications in the fields of entertainment, the media and advertising. International collaboration includes the visual effects (VFX) company Framestore, winner of three Oscars ; the developers of Brainstorm, Cubic Motion and IKinema technology; the largest certification authority in Europe, InfoCert; research groups at Pompeu Fabra University, Universität Augsburg and Inria; and CREATIVE WORKERS-CREATIEVE WERKERS, pioneers in immersive virtual reality.

This site uses cookies and analysis tools to improve the usability of the site. More information. |