Robot home brings together family of research

For the past decade it has been one of the undisputed research success stories of the University: a relatively small team, successful in bidding for multi-million pound grants, leading some of the biggest collaborations in the field, and establishing itself among the most respected peers in the sector.

From ITALK to ALIZ-E, the advancement of robotics at Plymouth has been anything but science fiction. 

Now with the creation of a new ‘robot home’ on campus, the work of the Centre for Robotics and Neural Systems (CRNS) is entering a new phase. Located on the first floor of the Portland Square Building, it is more than just an academic office with some IKEA furniture: it’s an interdisciplinary hub where an international team of researchers can study human interaction with the University’s ‘robot family’. 

“We have had three big waves of multiple project funding to get to this stage,” says Angelo Cangelosi, Professor in Artificial Intelligence and Cognition. “Robot Home could be considered a 2.0 or 3.0 for us because it marks an important change for the way we work. It signals a transformation from independent projects each focusing on a specific topic – such as iCub looking at language acquisition and manipulating objects, or ALIZ-E on social interaction and communication – to having a full-blown experimental setup, a living home for the elderly and robots.” 

A great example of this is a new EU Horizon 2020 project focused on supporting independent living through robotic companions. MoveCare: Multiple-actors Virtual Empathic Caregivers for the Elderly is being coordinated in Milan, with Plymouth leading a €440,000 strand on social interaction. Working with Professor Ray Jones, in the School of Nursing and Midwifery, as co-investigator, Angelo welcomes elderly people from a care home in the city to the centre, where they interact with ‘Pepper’, the newest addition to the family, courtesy of Japanese company Softbank. 

“We were among the first in Europe to receive a Pepper, and ‘she’ will be at the heart of a number of projects,” Angelo says. “In MoveCare, we will be able to study how people interact with her in a more realistic home-like scenario, rather than a laboratory with white walls. There is a great deal of interest in researching and developing robots to be used as domestic home helps for the elderly and the disabled.” 

MoveCare is one of nine current projects, representing an income to the University of around €4 million, and which have some 15 PhD students, and five postdoctoral researchers attached. Among that PhD population are seven funded by the Marie Curie scheme, each working on one of three projects: APRIL, SECURE and DCOMM, covering the topics of personal robotics, safe robot interaction, and communication using gestures, respectively. It is an indication of the importance being placed upon the subject of ‘collaborative intelligence’ – where humans and robots learn from one another to achieve a task.

“Industries are moving into AI and robotics, particularly in areas such as deep learning, which uses deep neural networks,” Professor Angelo Cangelosi says. “We have been working in that field for a number of years already so this gives us a significant advantage. Indeed, we have undoubtedly contributed to this growth.”

Professor Angelo Cangelosi

Robot learning has been at the heart of a number of Professor Tony Belpaeme’s projects over the past decade, and his latest, L2TOR is transplanting that principle to schools and language learning. The three-year Horizon 2020 initiative is focusing on teaching English to native speakers of Dutch, German and Turkish, but also teaching Dutch and German to children whose first language is Turkish. 

Using the Nao robots from the ALIZ-E project, children are invited to take part in a language game that involves learning numbers, spatial language and basic vocabulary through storytelling. Even in its early stages, L2TOR has found that the most cutting-edge speech recognition software cannot cope with the grammatical idiosyncrasies of children aged 4–6. It has also revealed a number of new avenues for enquiry, including the use of ‘deep learning’ to help the artificial intelligence overcome its inability to read emotion in faces. 

“As was the case with ALIZ-E, we are exploring the possibilities and the boundaries of AI and robotics,” says Tony. “It is not a straight line from A–B, but that is part of the challenge and excitement. We are still a number of years away from robots being able to interact with people on a deep level, but we’re making progress all of the time.” 

L2Tor is already generating huge interest in both the educational and robotics sectors, and entertainment giants Disney has even invited one of the project academics to work with its technologists in a bid to improve the experience of its theme parks. 

The presence of Disney – and the likes of Honda, and the US Airforce, with whom Angelo is working on a project called THRIVE, looking at the issue of trust with robots – is a very important one. At a time when Google has purchased the company DeepMind, commercial interest in robotics and AI is at an all-time high, which creates new avenues for funding. 

“One area that I think we can move into is human–car interaction,” Angelo confirms. “We have a growing relationship with Honda, so we have applications in the pipeline that focus on AI and cars.” 

With Brexit on the horizon, the change of funding landscape will be especially challenging for robotics research, and will call upon every competitive advantage we can muster. And that’s why Robot Home is an important asset in CRNS’s continued evolution. 

“Plymouth is now among the main players in the field, and we make a major contribution to the Engineering and Physical Sciences Research Council Robotics and Autonomous Systems network,” adds Angelo. “The difference between us and the likes of Bristol, Edinburgh and Sheffield, is that they can draw upon multiple universities to create a bigger critical mass. But with our links to industry (for example, strategic partnership with Aldebaran-SoftBank Robotics), and our focus upon interdisciplinary projects, we are able to run innovative studies on social and learning robots, create and test more realistic scenarios, and this helps us with future grant applications, impact case studies, and commercialisation opportunities."

Research - Centre for Robotics and Neural Systems (CRNS)

The centre builds on world-leading excellence in computer science, robotics and neural systems research. Staff at the centre coordinate large projects and collaborate with major international centres in cognitive robotics and computational neuroscience.

Centre for Robotics and Neural Systems (CRNS)