The nuts and bolts of new robots


Paro may be the most famous companion robot around today, but he’s likely to have some competition before too long. Here are a few others currently in the works:

The “emotion bear.” It laughs. It sneezes. It waves. It strikes up a conversation. And if nothing much is going on, it falls asleep. A concept currently being tested by Fujitsu in Japan, the “emotion bear” can sense when people are near and turn to face them. And when it gets to know people well enough, it can tell what mood they’re in — and behave accordingly. In tests at nursing homes and adult daycare centers, the bear was a big hit with the elderly, says Paul Moore, senior director of mobile product management for Fujitsu America in Sunnyvale. “When they saw it, their whole demeanor would change.” And preliminary evidence suggests that for some patients, just seeing the bear may increase blood flow to the brain — which could help ward off dementia.

Huggable. This adorable bear, under development at the MIT Media Lab, is designed to be a team player. In a healthcare setting, “Huggable” will join forces with a patient and a hospital or nursing home staff member. Whenever and however the patient interacts with the bear — tickles it, pokes it, slaps it, hugs it — it will respond appropriately. Meanwhile, it will be collecting behavioral data that the staff member can incorporate into the patient’s care.

Bandit. Some engineers try to make their robots as lifelike as possible, but not Maja Matari¿, director of the USC Robotics Research Lab. Studies have shown that people respond best to robots that are reasonably lifelike, but not overly so. “Bandit,” the robot used in most of Matari¿’s research, gives only a vague impression of being human, but it can be programmed to do lots of stuff, like leading people through an exercise routine. “I don’t care about robots just being entertaining — I want them to change people’s behavior,” Matari¿ says. Her lab is working on ways to use physiological data to interact with people more effectively. For instance, if the robot “coach” detects a change in how much someone is sweating and remembers that this has been associated with a tendency to quit exercising, Bandit can take a new tack to try to keep the person going.

—Karen Ravn