It turns out you can teach an old dog—or, at least, a middle-aged one—new tricks. Without any vocal or visual cues, a Labrador mix named Tai can turn, back up, lie down, and come on command. He does all this in response to remote-controlled vibrations in a newly developed dog vest designed by Israeli scientists at Ben-Gurion University of the Negev.
In results presented this week at the IEEE World Haptics Conference in Tokyo, the researchers report that cues issued by gentle vibration motors in the vest were as effective as vocal commands. Of course, Tai’s favourite treats of sausage pieces helped make him a fast learner.
Haptics refers to any technology involving the sense of touch, and there are practical and important applications for a haptic dog vest, explains lead author Yoav Golan, a PhD candidate in mechanical engineering at Ben-Gurion University. He has been Tai’s owner ever since the dog failed an exam to be a seeing-eye guide when he was a lovably hyperactive puppy.
“Communication with working dogs is still predominantly visual and audial,” the team writes in their paper. But search-and-rescue or tracker dogs often work at a distance from their handlers, under rubble, in small spaces, or out of sight and earshot. The haptic vest would allow people to issue instructions even under such circumstances. Vibration commands could also be suitable for secret operations done in silence.
A Basset hound.
PHOTOGRAPH BY JOEL SARTORE, NATIONAL GEOGRAPHIC PHOTO ARK
Similarly, police and military canines might find the vest useful if they are working in loud areas, such as conflict zones filled with gunfire, where vocal communication is difficult. Handlers and even dogs “sometimes use ear muffs to protect their hearing in noisy situations, such as aircraft travel,” Golan says.
In addition, haptic cues can be issued by anyone regardless of language. Tai, a six-year-old Labrador and German Shepherd mix, understands verbal commands in Hebrew, not English, but speakers of any language can issue a command with a remote control. Dogs also might be attuned to the voices of specific handlers or trainers, but vibration commands are likely agnostic.
“In my world of conservation dogs, the possibilities of using this vibration technology with tracking dogs has huge potential,” says Will Powell, a canine expert who trains sniffer dogs for African Wildlife Foundation and other anti-poaching programs in Africa.
“Dogs perceive their environment through their senses, and touch is a surprisingly important sense for a dog,” adds Powell, who was not involved in the new research. “Any working dog disciplines that involves distant work could benefit from this technology.”
For domestic use, the vest would allow pet owners to call and direct canines from afar. For example, if a dog is allowed to roam outside alone, an owner can push a button to tell the animal to return home. Vibrations would also give pet owners with speech impairments another way to communicate with their dogs, and the vest could help them communicate with a hearing-impaired dog, Golan says.
Consumer vibration dog collars do exist, but their signals are imprecise, and they are more often used as deterrent “shock collars” to curb behaviour such as barking or lunging, rather than teaching dogs to follow various, specific commands.
The new research instead expands on previous work evaluating the use of painless vibrations for dog training, including a 2017 paper that tested the effectiveness of haptic commands on dogs of varying sizes, breeds, and coat types.
For the latest study, the Israeli team built a special casing for the vibration motor that sits inside the vest. The rectangular casing is about 38 millimetres long, with elongated nubs that enable better contact with the skin through the dog’s fur. These nubs more effectively transmit vibrations to bony areas near the dog’s hips and shoulders and don’t require shaving his coat. Inspiration for this design came from coauthor Ben Serota, a neuroscientist who works with comb-shaped “dry electrodes” that attach to patients’ heads without shaving their hair.
Tai already knew the four vocal commands for turn, lie down, come, and back up, so teaching him those haptic cues was “not a large leap,” Golan says. The “smart, but not very smart” dog learned his first haptic command—to spin, or turn in a circle—in about an hour. His second command, to lie down, took longer to learn, partly because scientists had to adjust a motor on Tai’s hip so he could feel the vibration. A third cue, to come, took 15 minutes to learn, according to the paper.
Golan tested very low vibrations on Tai, and the dog did not respond at all—a sign that the pooch was not responding to human body language but to the haptics alone. Commands were also issued randomly by an unseen operator, so that the visible dog handler didn’t know exactly which cues were being issued but could observe Tai’s responses. (See how "puppy dog eyes" may have evolved so dogs could communicate with humans.)
The results showed that the dog could associate discrete commands with vibrations from four different motors near the top of each of his legs. The dog was also able to distinguish between steady vibrations and pulses, which shows the potential for dogs to learn commands of varying complexity.
"A truly good boy"
For this work, Tai has become Israel’s only government-registered research dog. Before any experiments, it took months to get approval from Ben-Gurion University’s animal research ethics committee and Israel’s ministry of health. The university also has veterinary staff who check on experiments and the conditions for animal subjects.
The haptic equipment, including the vest and motors, weighs about a pound—a negligible burden for Tai, who weighs roughly 35 kilograms. The experiments were non-invasive and painless; Golan tested the motors on himself first, and he says the vibrations are equivalent to the buzz of a cell phone. Tai did not show discomfort, had ample chance to rest, and received positive reinforcement as well as edible treats. In their paper, the research team credits Tai as a “truly good boy.”
Golan envisions additional research to test the haptics vest with different dogs of various sizes, breeds, ages, fur types, and training backgrounds. For example, would dogs with no prior training at all respond to haptic cues? How would they respond to completely new commands?
He hypothesises that dogs could one day be trained in remote two-way communications with handlers by biting on part of a vest when questioned by a haptic cue. Golan has also wondered whether a canine could carry a robotic arm on his back to do work remotely, thus combining the mobility of a dog with the precision of a machine.
Luckily, Tai is much beloved in Ben-Gurion University’s robotics lab, where he is an unofficial mascot. And the lab isn’t lacking for more potential canine subjects, Golan says.
“So many students want to bring their dogs.”