At home with the android family

At home with the android family

At home with the android family

On a weekday morning in a Hertfordshire street, people are knocking on the door of an ordinary-looking house. Inside, a living room hosts a sofa, bookshelves, coffee tables and a TV. Through an archway, the kitchen kettle is boiling up, ready for the first of many cups of tea.

So far, so normal –– but there’s something different about this home: it’s stuffed full of more technology than your average branch of PC World. Sprawled around its ground floor rooms are a family of robots belonging to the University of Hertfordshire’s school of computer science. This is probably the UK’s only robot home.


It’s part of a project that began in 2005, when Kerstin Dautenhahn, professor of artificial intelligence at Hertfordshire, was working on a European-wide piece of research called Cogniron. The aim was to create a “cognitive robot companion” for humans, and the team began building and modifying the machines. When they were ready for testing, the team invited people to their lab, where they were monitored while they interacted with the robots.

“It didn’t work well, because the participants didn’t feel very comfortable in such an artificial context,” Dautenhahn explains.  She decided to take the project out of a campus setting and into the home, so the academics could investigate how robots work as personal companions in one of mankind’s most natural environments. The robotics faculty first decamped from their laboratory into a local flat, but that soon became too small.

“So, in 2008, the university bought a two-storey house with a large ground floor area, so our robots and participants have a lot of space to move around,” Dautenhahn says. “All the furniture makes the house look comfortable, giving research participants the feeling of visiting a friend –– it’s not their home, but they could imagine living there.”

With the testers relaxed, the robotics team could carry out a range of experiments to develop their robots’ ability to work with – and for – humans. The projects differ depending on the particular issue the researchers are working on.

In one, a person sits at a writing table, triggering a robot to fetch a pen. In another, robots try to negotiate rooms without crashing into moving humans. A further trial programmed a robot to persistently interrupt TV-watching participants to ask if they wanted a diet Coke. If the tester said no, the robot repeatedly returned to offer alternative drinks, checking what kind of robotic interruptions participants would bear.

Now Dautenhahn is working on a “proxemics system”, controlling how close robots should get to people when approaching them. Earlier research found that humans felt alarmed when robots approached head-on, so the robots now approach from the side. Trials suggest that people are often happy for robots to get nearer than humans.


Like people, the robots vary: Dautenhahn’s arsenal includes human-sized mobile machines and a humanoid, toddler-like robot called Kaspar, whose rubber face and realistic features are reminiscent of the characters of the animation film “Up”. Dautenhahn is using Kaspar for the Aurora project, which looks at how robots can become therapeutic toys for children with autism.

“The children generally respond very well to the robots, playing with them, and exploring their abilities and physical characteristics,” says Dautenhahn. “Our goal is to help the children to interact and communicate with other people, so we’ve focused on using robots as social mediators: using a robot that encourages an autistic child to engage in interaction.” That, Dautenhahn says, is the ultimate target of all her social work on robots: she aims to develop machines that help people.

“It’s not about replacing people, it is about allowing robots to provide help in their homes. That’s especially important for elderly people – our work could allow them to stay in their own homes for longer.” To that end, the research team busily records all the robot-human interaction in the Hatfield house, with the academics watching participants from a small control room on the ground floor.

The tapes, plus the surveys the participants complete about their experience with the robots and their personal background, help the team work out how best to improve the robots and make them more like social animals. Although other researchers, mainly in Japan, focus on robotic engineering, the Hertfordshire work is distinctive in its focus on how robots can adapt to social behaviour.

Dautenhahn now has a team of 20 working with her, including PhD students and research assistants, with backgrounds ranging from robotics and engineering to psychology and computer science.

She thinks companion robots, with realistic human-like features and intelligent functions that allow them to speak and understand speech will be available within 100 years. She also expects robot girlfriends and boyfriends to be creatable, but worries of “a danger that people will then find it too hard to have real relationships, when it’s so much easier to have a robot that can be switched off when making annoying comments, and replaced so easily without arguments.”

In her own home, however, Dautenhahn gets a break from her metallic friends. “My house isn’t suitable for useful robots like robotic lawnmowers and vacuum cleaners,” she admits. “They can’t cope well with rooms cluttered with children’s toys and unusually shaped and uneven gardens, like mine.” And as for the worry that robots will take over the world, Dautenhahn thinks we can relax.

Get a round-up of the day's top stories in your inbox

Check out all newsletters

Get a round-up of the day's top stories in your inbox