Dr Tony Belpaeme has been awarded £165,000 by the EPSRC to study how robots can learn concepts from humans, and how that knowledge can be passed on to other robots.

Researchers at the University of Plymouth, are to build two robots that will learn the meanings of words through interacting with people, much in the same way that young children learn conceptual knowledge from hearing adults speak to them about objects, relations and actions.

It takes children three years to master a few hundred words and related concepts – the duration of this project, however, the researchers hope to speed up this process of word-concept learning by using training more than one robot, and so reducing the training time needed, and then downloading the missing knowledge from one robot to the other. Such ‘telepathic’ access to concepts is impossible for humans: we need to resort to pointing out examples of concepts and speaking about them, but direct transfer should be easy to arrange for robots. However, just copying information from one robot to another will almost certainly upset the conceptual knowledge already present in the receiving robot. To avoid this, direct transfer of conceptual knowledge needs to proceed with care in order to not disturb already present knowledge.

The project has two main aims.

The first is to study how a robot needs to behave in order to elicit conceptual knowledge from people. To do this the researchers will build a robot face containing cameras and microphones, on a long articulated neck. The neck allows the robot to look around the room, and for it to scrutinise objects laid out on a table in front of it. The robot will be able to seek eye contact, engage in joint attention and interpret gestures related to conceptual learning. It will engage in activities such as asking its human teacher to confirm a word or play a game of ‘spot the X’, to check its knowledge, and if necessary adapt it.

The second aim of the project is to design computer algorithms that effectively learn concepts from interaction involving real-world scene and words. Children are particularly good at this, and the reason for this is that they use a number of constraints to help their learning. It is these constraints that the researchers want to program into the robot learning mechanisms.