Research Magazine 2006

Intellectual robots

Engineers teach machines how to learn


Assistant Professor Manfred Huber and his research team are teaching this four-legged robot how to walk on its own.

Manfred Huber says robots can be incredibly stupid.

“You have to be very specific in what you tell them to do,” said the UT Arlington assistant professor of computer science and engineering. “For instance, a 2-year-old can understand what you want when you say, ‘Go sit in your chair.’ But for a robot, you must first define ‘sit’ and ‘chair.’ Now think about all the different designs of chairs. And a chair just isn’t someplace to sit; you can sit on lots of things. It’s a huge challenge.”

Dr. Huber is studying ways to improve human/robot interactions by giving robots adaptive learning skills—an artificial intelligence, if you will—that will enable them to grow intellectually from simple tasks to complicated actions and reactions.

“We’re developing mechanisms to give robots learning capabilities, to give them ways to look at the world, to know more about their environment and acquire skills,” he said. 

It’s one example of faculty and student participation in robotics research at UT Arlington.

Teams of engineering students are participating in competitions involving ground and aerial autonomous vehicles. For a challenge sponsored by the Institute of Electrical and Electronics Engineers, electrical engineering students Brian Hendryx and Reginald Golightly and computer science and engineering student Chris McMurrough are building a robot that can navigate a virtual warehouse. The moving machine identifies and gathers containers of different colors and places them in appropriate locations while avoiding obstacles.

Team members have spent about $500 out of their own pockets for parts and receive no course credit for their efforts. But they aren’t complaining.

“This is a good mix of engineering disciplines,” Golightly said. “For instance, we’ve run into a bunch of mechanical engineering problems to solve, and we’re developing time management skills. It also shows potential employers that we have a lot of initiative.”

Initiative took another UT Arlington student team to first place in an international unmanned aerial vehicle competition last summer. The project was a result of the students’ participation in the Autonomous Vehicle Laboratory, administered by faculty from several College of Engineering departments. The lab researches autonomous vehicle systems and supporting technologies.

The team created the first aircraft in the competition to complete a totally autonomous mission—takeoff, target acquisition and capture, and landing.

Robots that can learn have potentially far-reaching applications. In Japan, initiatives are under way to develop robots to assist the elderly and disabled. Japan’s aged population is getting larger and the birth rate is falling, so fewer people will be available to care for the elderly.

Humanoid robots are already in experimental use in Japan as greeters and in information booths. Consider, too, the benefits of a wheelchair that could, after a few learning experiences, go to a desired location with a simple request from its occupant.

“Machine learning and autonomous operation will make it happen,” Huber says. “It’s only a matter of when.”

— Roger Tuttle