Last Wednesday, I attended an evening at the ‘Secret Robot House’. I’m not sure what images that conjures up for you — an underground laboratory? White coats and flying sparks? Heath Robinson‘s spare room? A house made of robots, for robots, by robots? I wasn’t really sure what to expect. It was all in the spirit of the thing to not really know.
“A minibus will collect you from Hatfield station at 6:30 and escort you to the Secret Robot House”, was the only information from Alex (Internet of Things luminary and organiser of the evening). The minibus took us through a comfortable suburban estate and pulled up outside a residential house. A regular-looking semi-detached in a sleepy middle-class area of Hatfield. Alex waved us in and we entered.
It’s a real house, but not a real house. The first things I notice are cut-out spacial markers on the ceiling, dozens throughout the room. A desk covered in computers. Then a pair of robots by the bookshelf. One looks like an oversized Nespresso maker, the other Wall-E’s slightly junkier sibling. Everyone shuffles in an amused, but slightly confused, way. It’s a real house, but not a real house. It’s a robot’s home, a research space for domestic robots to understand human-robot relationships. We’re walking in the uncanny valley.
Nicolas Nova presented a short-talk that took us through the role of sci-fi in developing robots; robots being fictions from the very beginning, that people want to be real. We keep trying to make science fact out of science fiction. Nova talks so simply that it’s easy to imagine he’s not saying anything at all, but – and this is main the point of the whole evening – demonstrated how hard it is to design robots as robots. They will always be an expression of the designers’ beliefs, culture and understanding of others. Nova led us around the various ways in which people respond to robots and the ways that people don’t want to interact with robots (if they are humanoid, or animal-like; slide 11). It was an interesting wander over the developing rules of engagement of Human-Robot Interaction, what we are willing to interact with and what we are rationally uncomfortable about. Too human and we fear Nexus 6 Replicants; too much personality and we fear HAL 9000. And nobody really wants an internet fridge.
In response to Nova, Patrick Bergel suggested that robotics research owes more to old magic than it does sci-fi. He pointed towards the Sorcerer’s Apprentice as the defining example, an original manifestation of the desire for self-automated helpers to do our work for us. The brooms are robots in disguise. It’s true, inasmuch as the Golem is a clay robot, but it’s a little sophistic. We generally try to understand our futures by extrapolating upon the things that exist in our present. Magic and sci-fi are the same, except that magic is now electronic and cybernetic. Perhaps, if Walt Disney had been born thirty years later, the brooms would have been bots with screens.
Prof. Kerstin Dautenhahn (founder of the Adaptive Systems Group responsible for the robotics research in the house) gave us a deeper backstory to the research and the role of the house. She offered a better understanding of the limitations of robots and the perennial challenges that face the researchers. The research was about developing robotic companions — for the elderly, disabled or house-bound — to support and complete simple tasks. It is less about having fully functioning automatons that do everything, without input. Their challenge is to design and build Companions that are ‘useful’ and ‘social’, but also adhere to current and future standards of robotics.
An interesting point Dautenhahn made was about robotics (particularly the robotics research they are doing) being a synthetic science. It is a mixture of engineering, psychology and sociology. The robots aren’t a species, and are inconsistent. They change regularly with each new design/build. Intelligence is redefined with each iteration, emotionally and in processing ability. A robot of today is not the robot of the future, today. What the ASG are trying to do is develop a kind of social language, the laws of interaction between humans and robots beyond Asimov’s Three. How does a human react when a robot does this, and why? What is the best way for a robot to deliver a drink, how can a robot avoid injuring humans in delivering simple tasks?
Coming back to Nova’s talk — citing Kevin Slavin, “it becomes real by behaving real, by demonstrating the behaviour of things that are real” — we keep asking ‘how much of a human personality should you put into a robot?’
Over the course of two demonstrations, it became very clear that the role of human-ness in robotics is important. The Care-o-Bot 3® had two clear facets to its construction: a human-facing front, delivering objects and scanning the room; a mechanical reverse with massive arm for task-completion. The Care-o-Bot 3® moves like Mrs. Danvers in Rebecca, eerily smooth and slightly jarringly. It is neither wholly robotic nor wholly humanised. The uncanny valley is dipped into as the non-human movement sticks out more than it bleating “HERE IS YOUR DRINK” to complete its task. Communicating with human terms after decidedly non-human movements.
The second demonstration was of the Sunflower robot: a machine that definitely chimes with our expected ideas of a robot. Short, boxy, arms and a head with ‘eyes’. The eyes aren’t real, but are simply there for human benefit — to focus upon them, rather than looking for meaning in the rest of the machine. The Sunflower is the stranger of the two robots. It navigates a space, stops, recalibrates where it is. Occasionally it looks a bit lost and frightened, before it gets where it should be. It nags for attention when the doorbell rings, and is satiated when its master acknowledges it. It is, in the words of Bashford, a bit like Reality Clippy. Unlike Care-o-Bot, it uses flashing light systems to communicate if it needs attention or if a task is completed.
The other interesting concept at play in Sunflower is the ‘Migration’ mode, to make your robot portable. The Sunflower system is being designed as an “UbiCompanion“: the robot is not the machine, but the personality users have refined and customised. It is the functions that users have tweaked for their personal preferences. In Migration mode, users can transfer the robot’s consciousness, for wont of a better word, to other objects (think transferring a Mii character on your Wii to another console). This makes for a smarter system that jumps between a dumb-box on wheels, to skype, to desktop computers and anything that is connected. It’s a connected thing of the future, and very real.
The trip to the Secret Robot House was hugely enjoyable, and made for some interesting thoughts about behaviour and future-relationships. Huge thanks to Alex for organising and to LIREC for opening up their dizzying research. More organisations should invite plebs into their labs.