Robots Get a Light Touch
Professor Peter Whitney shows a safer robot is possible.
It’s the Holy Grail in robotics: an android that can perform complex tasks beside humans in a real-world environment.
The problem is that when robots are asked to operate outside a highly controlled environment, they tend to be slow and clumsy. The fast and precise automations that work so effectively in factories can’t work in real-world environments because they can’t immediately sense and respond to their environment. Ask them to quickly grab a fragile object, and they might crush it. Ask them to work with a human co-worker, and they might accidentally knock his head off.
“If we want robots to work alongside humans or handle complex tasks autonomously, we have to make a gentler robot,” says Peter Whitney, an engineering professor at Northeastern who specializes in robot mechanics.
Whitney believes he has the answer: hydraulics.
While working at Disney Research in Pittsburgh, he developed Jimmy, a remarkably lifelike robot that’s fast, precise, and gentle enough to play catch with a balloon, pick up an egg, and pat a girl’s cheek without giving her a concussion.
For now, Jimmy is operated like a puppet by a human. This early incarnation demonstrates that Whitney’s hydraulic system can create a sense of touch that vastly expands the robot’s capabilities. It doesn’t crush the egg because the operator can feel its shell through hydraulic feedback; it doesn’t clobber the girl because the operator can feel her face as soon as the robot arm touches her.
Even though Jimmy is not yet autonomous, the puppetlike system has many applications. For example, Whitney is working with researchers from Stanford University to adapt his technology to perform robot-assisted MRI surgery.
Because an MRI tube is so narrow, doctors currently have to roll patients in and out of the tube as they perform surgery. Whitney is working to perfect a smaller version of Jimmy to perform surgery inside the tube, guided by the surgeon using the puppetlike controls. Because the hydraulic technology transmits a sense of touch, the surgeon can actually feel what Jimmy is doing.
Whitney speculates that if the key parts of the hydraulic system can be miniaturized, it might be adaptable to laparoscopic surgery.
“Laparoscopic surgery is now completely numb,” says Whitney. “All of the surgeon’s movements are controlled by sight, nothing by feel. That’s because there’s no accurate force feedback.”
The next step, he says, is to use his system in autonomous robots, combining the speed and precision of industrial robots with the gentleness needed to work alongside humans. The potential applications are enormous, ranging from eldercare and rescue work, to hazardous waste removal and space exploration. Whitney says the target for this technology is any work that requires a robot to be precise and fast in one moment and compliant and gentle in the next.
Less than 5% of occupations consist of activities that are 100% automatable.—McKinsey, A Future that Works, 2017
Whitney is working in collaboration with Northeastern computer science professor Robert Platt, who is developing the artificial intelligence needed to perform autonomous tasks—particularly grasping and manipulating common household objects so a robot can be used for tasks such as feeding and shaving.
“With Jimmy, we demonstrated what the hardware will be physically capable of when the artificial intelligence catches up,” says Whitney. “Even though it’s still human controlled, it proves that an autonomous application is possible.”
Whitney’s hydraulic system also has the potential to dramatically reduce the cost of agile robots. This is particularly important in the home setting, where customers don’t have the financial resources of manufacturing companies.
“Our goal right now,” says Whitney, “is to develop a robot arm with at least seven joints, that can be mounted to a wheelchair or mobile robot-wheeled base, with a total system cost under $2000, which could be cut in half through mass production.”