The Future of Robotics

This story originally appeared on page 64 in the Spring 2011 issue of Illinois Alumni. Seth Hutchinson is a researcher in the Coordinated Science Laboratory.

Dec. 10, 2010 -- The Associated Press reports a standoff between police and the driver of a stolen car parked along a Milwaukee freeway. Police send a robot to break out the car's rear window, allowing officers to throw in tear-gas canisters from a safe distance.

Dec. 26, 2010 -- Roomba™ Robotic vacuum cleaners, unwrapped as gifts, hum around living rooms, skirting objects before parking themselves in a charger.

Robots have arrived in our daily lives.

These are "not very interesting robots," though, by the standards of Seth Hutchinson, a leader of robotics research at the University of Illinois. Hutchinson is director of CAESAR (the Center for Autonomous Engineering Systems and Robotics) at Illinois' Information Trust Institute. He is working toward the day when robotos resemble the walking, talking androids we know from movies.

Androids are still decades away, but many new applications of robotic technologies are being developed in which robots and humans share responsibility. Consider the drone aircraft.

"Drones are semi-autonomous vehicles carrying out tasks -- robots, but they don't understand the world," Hutchinson says. "A human operator somewhere understands the situation. So you decouple the problem into parts that robots are good at -- taking incoming fire and delivering ordnance in places that are difficult to reach -- but relieve the robot of deciding who is a terrorist and who is a civilian."

Surgeons employed robots in 86 percent of prostate surgeries in 2009, using tiny instruments to do very precise work, and Hutchinson expects much of the imminent growth of robotics to be in the health-care industry. Part of the reason is robots' surgical precision, but simple economics also plays a role. With a large, aging population and dwindling numbers of caregivers, Japan is trying to build caregiver robots that could "help Grandpa get up if he falls down, give him medicines, test to be sure he's taken his medicines -- things you'd want a nurse to do," Hutchinson says.

But if robots are to gain that level of autonomy, significant issues of trust must be overcome. Advances in haptic technology geared toward tactile issues would allow more human-robot interaction through touch. (Humans have to trust that the robot knows how much pressure to apply when it grabs Grandpa's arm, for example.)

Huthinson says the big challenge is to build robots that understand the world around them. "For robots to decide the difference between an ottoman and a baby on the floor -- that's something they don't do well," he says.

Yet.

Solutions are coming faster because of the exponential growth in computing power. According to Hutchinson, "Computer science is now saying, 'You hit me with all the data you've got, and I'll find patterns in it.'"