Robots: Learning to Learn from Their Environments

University College Dublin are to be involved in a European collaborative project which will see the fostering of self-learning robot ecologies. These environments, in which robots will learn from each other and work together to find efficient solutions to various problems, are part of the RUBICON (Robotic UBIquitous COgnitive Network) project.

Mauro Dragone of the CLARITY Centre, a partnership between University College Dublin, Dublin City University and Tyndall National Institute, Cork. is the scientific leader of this project. He explains that, “The concept is that robotic devices are becoming more diffuse in everyone’s home, and we see value in making those appliances, those sensors, actuators, mobile robots work together”.

Mauro emphasises that they have a broad view of the idea of what a robot is, “We include, of course, mobile robots, but also static sensors or actuators, even automated home appliances.

“If these robots work on their own we don’t see much value, but if they work in a co-ordinated fashion they are able to achieve services that are useful in a variety of applications.”

Healthcare is a particular area of emphasis for the project, but Mauro hopes that RUBICON will see even more practical applications in the home.

“Think about service robots that help the elderly to live more autonomously in their own home but also robots in industry. For example, imagine the case of an automatic vacuum cleaner that would avoid cleaning when any of the inhabitants are home after receiving information from the home alarm system. If the home alarm system co-operates with the automatic vacuum cleaner they provide a service that is of more value than just an automatic vacuum cleaner and a home alarm system as independent entities.”

With many robotics projects, the expense involved outweighs the practical benefits of the technology. Mauro points out that, “It is the need for human configuration, for human programming that is impacting on the cost of these solutions. But if they can learn automatically, autonomously, they will be much cheaper to deploy in a variety of applications.

“Think about the Rubicon project as being something like the movie Avatar. We’ll achieve something like that for robots. Robots will be able to share their knowledge, not only by exchanging data but also by learning in a distributive fashion. How we achieve this is by creating distributive neural networks.

“A robot will enter this Rubicon, this intelligent robotic ecology, by sharing wireless connections with the other entities in this ecology. Once a device learns something they learn to operate in a co-operative fashion and they support each other’s learning.

“The idea is that we deploy one of these systems in a specific application, and the system will learn to adapt to the environment and to the application, basically learning to operate more efficiently without requiring human supervision, which will be one of the main achievements that we’ll target in this project.”

The idea of robots learning independently has been the subject of science fiction for decades, and the source of much paranoia, typified in films like 2001: A Space Odyssey, The Stepford Wives and The Terminator. The RUBICON projects are not, notes Mauro, that far advanced. They are, he says, ‘goal oriented.’

“Basically these ecologies will have a specific goal, so they are deployed to solve a problem, to help the elderly, to transport medicines from a room to another room in the hospital. Actually, a transporter application is one of the tests where we will evaluate our technology.

“They will learn how to do their job, the job they are programmed for, but they will learn to do it more efficiently, in an independent, autonomous fashion.

“They still achieve the job they were programmed to do, but the fact is, they will do it faster, they will do it better, and they will do it in a co-ordinated fashion without interfering with other human activities.

“They will do what they are told to do, but they will do it better without requiring constant human supervision and programming. That’s the main target of the project. So we don’t see them as taking over any time soon, you know?”

RUBICON is a three-year project, funded by the European Commission, commencing in April 2011.
 Mauro Dragone is a co-principal investigator, project manager and team leader at the CLARITY Centre for SensorWeb Technologies in Ireland and a postdoctoral researcher at UCD’s School of Computer Science and Informatics.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s