We are creating an artificial intelligence that can learn socio-communicative skills by observing and imitating people in an interview.

Most people spend countless hours using their communications skills – practicing them “on the job” every day. Since training opportunities for such skills are in ample supply in modern society, they are a prime target for imitation learning.

Robots that are intended to collaborate with people could benefit greatly from applying imitation for learning social skills. Since multimodal socio-communicative skills are a kind of meta-interface, that enables other kinds of learning to happen through interaction with (human or robot) teachers, this is even more motivation for endowing artificial agents with communication skills.

For these reasons (and many others) imitation learning of socio-communicative skills is an extremely interesting and important research topic with significant potential benefits for the robots of the future. more->

The HUMANOBS project started in January 2009 and ran until June 2012. Software, papers, documentation and other relevant information has been made available on this website – and will be updated regularly over the coming two years as more is produced.