Barking byte: man's new best friend

Artificial intelligence just got smarter: it developed artificial emotions. We look at the pervasion of personal robotica in the form of the pet robot - and it really is good for your health...

  • E-Mail
By  Kate Concannon Published  June 5, 2002

Introduction|~||~||~|“I called to Bob again and again,” read an entry stored on one AIBO’s personal daily diary after several days of sitting on the back burner while its owner attended to pressing work matters. “When I read it, I almost cried,” said Bob — big, brawny, scientific kind of guy and loving AIBO owner. Such is the response to increasingly sophisticated inorganic cuties: personal robots who flaunt artificial emotion and intelligence to distraction. CHARGED talked to Professor Takanori Shibata, senior research scientist for Japan’s National Institute of Advanced Industrial Science and Technology, and creator of Paro, the seal robot, to contemplate the bold new horizons this field of robotics is unzipping for the human experience.

When people first conceived of robots, they envisaged them as machines that would relieve humans of unpleasant chores, freeing up our time for more interesting ‘human-worthy’ pursuits. Aside from the spate of vacuum cleaning and lawn-mowing robots that have appeared — enjoying less than screaming success — we have seen very little of this vision delivered to real life.

Elektro, a walking, talking, even smoking robot (complete with canine robo-pet Sparko), was presented to the world by his Westinghouse creators at the 1939 New York World’s Fair. In the sixty-odd years that have since transpired, robots have revolutionized industry. And yet the sophisticated android that would provide hi-fidelity companionship or throw the drudgery of household chores into the realm of history has remained at the distance of sci-fi imaginings.

Where the robot has shown consumers promising performance to date is not in clunky, anthropomorphized machinery but in a smaller, cuter — albeit less ambitious — form, the pet-o-morphic personal robot. Think AIBO, RoboScience’s RS-01, Poo-Chi and I-cybie.

It is these entertainment robots that have won genuine consumer appeal and enthusiasm. This is the space of economic viability — consider the initial release of Sony’s first AIBO: when first put up for sale in June 1999, 5,000 units were sold online in just minutes, despite their $2,500 price tag.

Contemplate the expectations that consumers today would have of an android. Growing up with the Jetsons, A Small Wonder and then treated to films such as AI, our understanding of what such a robot should be willing and able to do for its keeper are vastly disparate from the reality of technology at hand today.

“We expect something that looks like a human to be capable of all of which a human is capable,” says Professor Takanori Shibata. Herein lies the crucial difference that has accommodated the success of animoid robots.

Who expects a cat to understand the concept of pointing (give up if you’re still trying)? Who expects a dog to initiate witty and flattering conversation while simultaneously scrubbing down the bath-scum and pouring drinks? Our expectations of an animal’s ability to interact with us are far more compatible with the current standard of AI and mechatronics, and so we are able to accept them far more easily.

At least this is the premise upon which Shibata has based his pet project, Paro, the seal robot. “People expect an android to be mobile, communicative, emotionally responsive and to have a somewhat independent mind. At this point — and certainly this was the case at the time when I did my preparatory research into the matter in the early 90s — the technology is not sufficient for a creator to produce a humanoid that meets these expectations.”

The technology does exist, however, to produce behaviours that reasonably resemble animal behaviour. This is key to paving the way for man-machine interaction. If we can accept a robotic animal, then we are in a far better position to deal with other machines, which will, in years to come, become more predominant and important in our lives. Robots will increasingly deliver all manner of services, and provide care and entertainment. “Mental commit robots,” as Shibata refers to this emerging breed of AI creatures, “shows a whole new role for robots in human lives.”

||**||Animal assisted activity|~||~||~|But it’s more than just that. Animals affect us in very real ways, mostly positive. The ultimate dispensers of unconditional love, they offer unique therapeutic benefits for humans. Studies have long indicated a positive relationship between convalescence time and animal companionship, which seems to reflect in turn the connection between psychological and physiological well-being.

It is this very human benefit Shibata decided to explore with his Paro project. Allergies, hygiene, tenancy agreements and housing space are factors that greatly impact the feasibility of pet ownership by an individual. The risk of selecting a pet that goes on to demonstrate threatening or antisocial behaviours is also a deterrent for some potential animal keepers.

But Shibata believes that everyone can and should benefit from interaction, and robot pets bypass many of these impediments. “Robots are not about replacing animals, but about complementing them and bringing the benefits of animal company to those of us who cannot keep them. Some people, especially in hospital and nursing home environments, have allergies and are afraid of infection and bites. Robot pets can help those a real pet cannot.”

Paro represents a nine-year opus undertaken by Shibata to develop a robot, which brings healing and therapy, as well as entertainment, to interaction between robot and humans. Shibata has been producing a revised version of Paro each year — though Shibata explains that his relationship with Paro through each incarnation has progressed as though it is the one entity.

The current Paro has a 32-bit RISC chip; tactile sensors on the whiskers and running along most of Paro’s body; posture sensors; two light sensors; seven motors; two microphones for sound localisation; and seven degrees of freedom.

When Paro is revamped the next time around, the actual build will be more modular to facilitate commercial production (Paro should be on shelves in Japan by the year end, and available elsewhere in 2003), and Paro will also be able to move its mouth. This will allow it to ‘suck’ on its recharge dummy for added effect.

To have the opportunity to play with Paro is a rewarding pleasure, to say the very least. Paro bats its eyelids, curls up to you, makes endearing noises, cooing and crying, and generally responds to stimulation (petting, noise etc) in a manner that seems completely natural and appropriate.

Paro generates behaviour and acts based on three factors: stimulation (frequency, type etc); the time of day; and internal ‘moods’, as Shibata describes them. There are lots of pre-programmed algorithms at work, but Paro is also capable of learning by reinforcement and is so well developed that behaviour appears entirely unprogrammed — it feels spontaneous but contextually appropriate to the extent that Paro, for all purposes, cradles in ones arms as an organic little creature with a mind and life of its own.

||**||Artifical emotion|~||~||~|He has taken his seal to the streets, so to speak, with extraordinary results. In one Robot Assisted Therapy (RAT) experiment, young hospital patients were allowed to play with Paro at key times of restlessness — before meals and in the evening. Nurses reported a reduction in patient anxiety and increased communication between patients and caregivers.

The Tsubuka University Hospital recorded particularly interesting results with a young autistic patient, who recovered his speech and appetite during the period in which the robot was at the hospital.

In another experiment, elderly people in a day care centre that played host to Paro for five weeks stated that they felt interaction with the robot both motivated them and promoted social communication. Stress among senior citizens and their caretakers was notably reduced, and nurses considered the robot assisted activity helped minimise burn out among them. Thus merit was threefold: psychological, physiological and sociological.

But why do we gush in suspended disbelief before these creatures when we know they are nothing more than a bundling of algorithms, silicon and shiny casing? “We are programmed to respond with affection to creatures that ask for our nurturance,” says Sherry Turkle, professor of the sociology of science at MIT.

“These new objects do something none of the earlier computer systems did: they ask for our care. When you give it to them, they thrive. It’s a feedback loop to an area in which we are emotionally vulnerable. These toys are plugging into that impulse. They’re pushing those Darwinian buttons.”

And it’s not just teddy bear owners and mushy types falling under the emotional spell. Many of the big buyers in this segment are hardcore techies — more traditionally expected to be found delighting in threads of code than bursts of artificial emotion emanating from a clever animoid. What begins here as technical curiosity or an intellectual exercise often ends in ‘love’.

As Shibata sees it, a robot’s overall subjective value and the effect of its interactivity on humans is equal to more than the sum of its parts. Breaking down the specific capabilities of Paro gives no indication as to the overall experience its behaviours generate.

They are consistent enough to be logical and appropriate to circumstances, but they are not routine and ritualised, creating a sense of highly natural and lifelike interaction. “Nor is subjective evaluation by people who experience Paro [in all its various incarnations] proportional to levels of technical specifications,” says Shibata. People are engaged and impressed by a machine that responds to them in natural ways that are immediately meaningful within the context of human experience.

||**||Animoid appeal|~||~||~|But even animoids are burdened with expectations, according to Shibata. People have preconceived ideas about exactly how a cat or dog should behave. Shibata has developed a number of animoids through the course of his career, including Tama, which he developed for Omron Corpora- tion who commercialised the product as NeCoRo.

NeCoRo has not proved as popular as might have been anticipated, however, for two seemingly conflicting reasons. “Some people find it creepy — they feel it is too realistic in the way it moves and feels to the touch.” Others feel it is not cat-like enough, and for that reason is disappointing.

Shibata considers that this a one of the very reasons that Paro has been so successful in attaining the acceptance of people from all walks of life. Who has pre-conceived notions about how a seal should interact with humans?

How easily and to what extent people accept Paro is a question of much interest to Shibata. Understanding how and why people relate to and interact with his robot over the years has given him vital information about what people require in order to be able to accept a machine and interact with it in meaningful, effective ways. “I hope my work help people to accept machines and robots like Paro not as tools or slaves, but as friends or companions in their lives.”

Culture plays a significant role in this. Despite some anxiety about how people would respond to Paro during its appearance in London, Shibata found that people were even more accepting of Paro than in Japan, where denizens, known for their tech-savvy and interest in all things electronic, reacted enthusiastically to Paro.

He believes this is related to the pet-loving culture of the Western. “In England and culturally similar places, people regard animals highly and value their pets as part of the family, so they make the desired connection very easily.”

This is not strictly the case with AIBO, on the other hand. While many individuals who are comfortable with technology derive emotional satisfaction from a relationship with AIBO, others are able to respect it for its technology without being able to make an emotional connection.

“Some people who find AIBO evil — although this is an extreme reaction in itself — feel that Paro is okay,” said Shibata. Why Paro is able to reach out and touch a wider spectrum of people is rooted in the fact that anyone can approach Paro and interact effectively.

An individual who struggles to program the VCR, let alone configure a PC, is bound to have a hard time getting through to AIBO, and certainly cannot access the wealth of emotional opportunity that exists with AIBO if you can get past operating it.

The vast numbers of AIBO enthusiast websites nonetheless attest to the satiating intercourse that is in fact attainable — as does the countless number of techies who have fallen under the spell of its artificial emotion at least as much as they are engaged by the power of its artifical intelligence.

As Sony marketing strategies rightfully indicate, when it comes to AIBO it’s not so much a matter of purchase, but adoption.

Add a Comment

Your display name This field is mandatory

Your e-mail address This field is mandatory (Your e-mail address won't be published)

Security code