杏吧原创

Dogs that help humans in danger can do so much more when they have technology designed especially for paws and noses

CLARA MANCINI鈥橲 colleagues used to laugh when she talked about her work. Mancini finds ways to improve our experience of interactive technologies. Should there be buttons? How many, how large, what colour? As gadget designers well know, little things can make a big difference.

But had shifted her focus to a group of users with very different needs. Dogs. 鈥淚t was very hard at the beginning,鈥� says Mancini. 鈥淧eople saw this as something not quite to be taken seriously, as something a little bit funny and cute.鈥�

Yet dogs are an integral part of our everyday lives, and that creates a growing need for them to interact with technology. 鈥淒ogs are already used in search and rescue, in medicine, as service animals, to help autistic kids and more,鈥� says , an electrical and computer engineer at North Carolina State University in Raleigh who is building a smart harness for dogs. He thinks technology can help rather than hinder. 鈥淲e鈥檒l be able to make them even better at their jobs,鈥� he says.

Gadgets for animals are on the rise (Image: Jonathan Burton)

But how do you design tech specifically for an animal? What are their needs? How will they best interact with devices? As research tackles these questions, we are learning how to help them help us better. What鈥檚 more, insights from working with animals may expand our horizons when it comes to building interfaces for ourselves.

Dogs might chase cars or howl at the TV, but how else do they interact with technology? 鈥淚 imagined seeing my dog sitting at a keyboard trying to type,鈥� says , a software engineer at the Open University, UK, where Mancini is also based.

That image may seem bizarre at first, but it鈥檚 actually not a bad approximation of the challenges many dogs face in performing the tasks we set them. Assistance dogs, for example, are trained to operate light switches or raise the alarm if their diabetic owner collapses. 鈥淲hat changed my mind was hearing about the ways humans ask animals to interact with technology, and then reflecting on how difficult and stressful it must be for them,鈥� says Sharp.

We have long been strapping gadgets to animals. Zoologists routinely attach GPS trackers to beasts of the land, air and sea to monitor their movements. The US military has even stuck sensing devices to dolphins and sea lions trained to detect underwater mines.

Crucially, however, none of this technology was designed with the animals鈥� experience in mind. 鈥淯sers need to be involved in the design process so researchers can know what they really want,鈥� says Mancini. That鈥檚 usually when the jokes begin. 鈥淧eople would say things like, 鈥楬ow would you get informed consent from a tiger? If it doesn鈥檛 bite your head off, does that mean it approves?'鈥�

But those turn out to be good questions. How do you design devices for a user that can鈥檛 tell you what it wants? Mancini鈥檚 team decided to focus their work on dogs, since they are already thoroughly integrated into our lives. Looking at the animals鈥� anatomy and behaviour is a good starting point.

Nose for precision

When Mancini began trials involving turning on a light, the dogs jumped time and time again for the switch, clumsily pawing at it but often failing to activate it. One expert trainer told her that she could count on dogs not being able to turn on the light even after several attempts. 鈥淪ervice dogs are struggling to do the tasks that are required of them,鈥� says Mancini. 鈥淔or them, operating a tiny, fiddly light switch is very difficult.鈥�

Dogs cannot easily distinguish the colour red either, so red alarm buttons and lights don鈥檛 register. Dogs also categorise objects by size more so than shape, and prefer to use their noses for precision tasks such as hitting a button or turning off a switch.

Taking that into account, the team designed dog-friendly buttons in different sizes and in high-contrast blue and yellow, each activated with the touch of a nose or the slap of a paw. When Mancini presented dogs with them, they were able to switch on the light at the first try. Such a simple change can make a dog鈥檚 life a lot easier. 鈥淚magine if you had to use a computer with really teeny-tiny keys and a tiny screen, and that every time you needed to turn on a light you had to jump,鈥� she says. 鈥淭hat would stress you very quickly.鈥�

鈥淚magine if you had to jump every time you turned on a light. It would stress you鈥�

In a second project, Mancini and graduate student Charlotte Robinson addressed the lack of any system for medical-assistance dogs to sound the alarm if their diabetic owner should suddenly collapse because their blood sugar has dropped abnormally low. The prototype alarm is a multicoloured rope that, when pulled with the mouth, sends a message to emergency services.

In July, Mancini and her colleagues exhibited their work at the Royal Society鈥檚 Summer Science Exhibition in London. A couple of wheelchair-using visitors who brought along their service dogs were delighted when their companions figured out how to work Mancini鈥檚 interfaces on the spot.

Well-designed devices can help dogs communicate in other ways. Mancini is also working with cancer-detecting dogs, trained to use their sense of smell. Normally, the dogs sit or lie down to indicate the presence of cancer cells in a breath or urine sample, for example, and return to their owners if the sample is negative. However, if dogs are unsure whether what they are sniffing is malignant, they sometimes turn to their owners for help, potentially biasing the evaluation. Like humans, dogs vary in their interpretation of evidence, with some animals more prone to giving false positives than others.

To get around this, Mancini and her colleagues designed a pressure plate that allows dogs to signal their degree of certainty according to how firmly they push on it. The devices are already being used by Medical Detection Dogs, a UK charity that trains dogs to do this work.

Others have gone further, designing what almost amounts to a doggy walkie-talkie. A team at the Georgia Institute of Technology in Atlanta, including , a key developer behind Google鈥檚 Project Glass, are perfecting an interactive vest for service dogs as part of a project called FIDO 鈥� Facilitating Interactions for Dogs with Occupations. The researchers consulted experts who work with police and military dogs about the tasks their animals perform and the types of clothing or gear that the dogs typically wear on the job. They then created several different sensors based on dogs鈥� ability to bite, tug, nose-tap and grip things with their mouths. 鈥淭he dogs know how to use FIDO vests for accomplishing a variety of tasks, just like a human knows how to use a cellphone,鈥� says , the project director.

FIDO allows dogs to communicate specific messages to owners or handlers in a variety of situations. A service dog, for example, can alert its hearing-impaired owner if a tornado siren goes off or the doorbell rings. A rescue dog that locates a lost child could stay with them and call for help by activating a GPS-tagged sensor, rather than running back to its handler. And a bomb-detecting sniffer dog could indicate what type of explosive it has found, and where, then get clear of the area. 鈥淔IDO is a game changer for working dog teams,鈥� Jackson says. 鈥淭hese highly trained dogs have information that currently they have no way to impart to their handlers.鈥�

鈥淲orking dogs currently have information they can鈥檛 impart to their handlers鈥�

Give the dog a phone

The FIDO vests contain nine sensors, each of which conveys a different message. Vibrating motors inside the vest can be used to give dogs commands remotely. The researchers tested various breeds, including border collies, golden retrievers, pit bulls, labradors and even Jackson鈥檚 own little papillon. 鈥淲e haven鈥檛 encountered a single dog yet that could not activate all of the sensors,鈥� she says.

Jackson designs the gear to ensure the dogs can operate it as easily as possible. 鈥淲ould you design a button for a cellphone that had to be bitten to operate it? Of course not, because we humans have fingers and thumbs,鈥� she says. 鈥淟ikewise, we wouldn鈥檛 expect a dog to be able to operate a button on a cellphone designed for humans.鈥� They also recently began collaborating with Georgia Tech鈥檚 K9 bomb sniffer-dog unit. The ensuing trials taught them that they needed to make the vests waterproof. 鈥淭he first time we tested it, the dog went straight into a lake,鈥� Jackson says.

Meanwhile, at the University of North Carolina, Jeff Bozkurt is working on a smart harness that will let dogs collect and transmit data from disaster zones. He thinks such gadgets are just the start. 鈥淲e鈥檒l find new applications for how these animals can help us,鈥� he says.

One such application involves teaming dogs with drones. Anna Gergely at E枚tv枚s Lor谩nd University in Budapest, Hungary, is working on a project called Swarmix, which trains dogs to interact with drones in search and rescue. The aim is to help rescue dogs find survivors as quickly as possible, the idea being that dogs can move much quicker than a human handler, while drones can gather a bird鈥檚-eye view of the scene. If the drone detects a victim, it will guide the dog to the spot.

Gergely is learning how to get dogs to work with robots rather than humans or other animals. She put food in a sealed box, which the dogs tried and failed to open. She then introduced a remote-controlled car that opened the box, allowing the dogs to get at the food. Eventually, the dogs began responding to the car as if it were a human handler. Finally, she began using drones in place of the cars. Although the dogs had trouble following high-speed Swinglet drones 鈥� a type of surveillance drone with swept-back wings 鈥� they adapted well to slower Quadrocopters.

Because they are easy to train, dogs are the obvious starting point. But technology could also provide distraction for bored animals on farms and in zoos (see 鈥�Poke to play鈥�). Jackson imagines that horses could be another good candidate. 鈥淐ertainly technology could be used to inform riders if their position is correct, of how much pressure they have on the reins and where their weight rests,鈥� she says.

Ultimately, though, such work with animals could teach us about our own interaction with technology. Are screens the best we can do? Discarding some of our assumptions could lead to better interfaces 鈥� ones based on tactile feedback, for example. Coming up with ways to interact with technology could also help us build devices for pre-verbal infants or cognitively impaired people. Researchers have experimented with different interfaces for tablets used by orang-utans and bonobos, for example, and their insights could improve interfaces for children with dyslexia or impaired vision.

More generally, tech designers have to come up with more ways to get user feedback when the user is an animal 鈥� using biometric measures or behavioural studies, for example. Similar techniques could equally be used to evaluate technology intended to be used by people who have difficulty communicating.

For now, Gergely is already seeing ways technology enhances people鈥檚 relationships with their animals. The owner of a dog that recently took part in one of her robot-interaction trials warned Gergely that the dog was stupid. But the dog ended up doing better than others that had been fully trained. The owner was overjoyed and enrolled her dog in training school. 鈥淣ow they take walks and have fun together,鈥� Gergely says. 鈥淚n that dog鈥檚 life, at least, there was a big change.鈥�

This article appeared in print under the headline 鈥淟assie gets an upgrade鈥�