OU Robotics: Machines learning real-world problems

Robots doing tasks for someone with disabilities or managing large amounts of information for medical researchers? The stuff of sci-fi novels and the distant future? Maybe not so distant, students and faculty said last week at the University of Oklahoma.

"What we are trying is to make robots cooperate with humans in daily activities," said Rafael Fierro, electrical engineering professor from the University of New Mexico who was a faculty mentor this summer for Research Experiences for Undergraduates: Integrated Machine Learning Systems.

Four OU students and six others from around the country participated in REU this summer, which took place at OU and the University of New Mexico. For about 12 weeks, the students were paired with faculty and professional researchers to conduct research.

The students presented the results of several research projects focusing on machine learning Thursday at Stephenson Research and Technology Center.

The program focused on applying machine learning to real-world problems. "Machine learning" allows computers to discover automatically how to pull out important information from complex data sets and how to perform tasks optimally. Machine learning can be used in robot control, severe weather prediction, computer security, brain-machine interfaces, computational neuroscience, bioinformatics, law and interactive art.

The art component might seem out of place, but it's a natural fit for Sam Bleckley.

A sophomore electrical engineering and fine arts major at OU, he was able to use both his studies in his project with Dr. Adam Brown, Dr. Andrew Fagg and other REU students. The team is developing the Orgonome, a three-legged robot that can learn from its interactions with humans. The long-range goal is to create dozens of these robots and then install them in an art gallery. There, the robots would interact with visitors and the display would get more and more interesting as they learned more human traits.

"Everything points toward the project being perfectly viable," Bleckley said.

OU senior Josh Southerland spent his third year in the REU program working on robotic manipulation. He focused his research on getting robots to complete the tasks of moving things around, with the hopes of providing helpful robots to people with disabilities in the future.

"What we're trying to do is find a balance between the robot completely controlling itself and us completely controlling the robot," Southerland said.

The REU program at OU is in its fourth year. It is one of many throughout the country funded by the National Science Foundation. The program was started after industry leaders like Intel and Microsoft told universities there is a serious lack of U.S. researchers with advanced degrees in computer science, said Fagg, the OU associate professor of computer science who helped write the grant to get the REU program at OU. So the companies are forced to go outside the U.S. for those workers, he said.

"So it's important for us to be building a pipeline into OU, but also in general, this is a way for us to be getting more undergraduates into computer science graduate programs," Fagg said.

The REU program gives undergraduates a taste of the research they would conduct in graduate school.

UNM computer science faculty mentor Terran Lane said he wishes he'd had such an opportunity when he was an undergraduate. He hopes he can help students catch the research bug.

"(To) get across the excitement of creating things that the world has never seen or discover things the world has never known."