Elderly people with limited mobility may soon come to be served by robots in a manner as if they are being served by other persons, thanks to a collaborative study by three University of Illinois at Chicago engineers and a Rush University nursing specialist.

"We want to help elderly people communicate with robots, to tell them what they need, and to perform physical activities," said Milos Zefran, UIC associate professor of electrical and computer engineering.

The three-year study, supported by a grant of 989,000 dollars from the National Science Foundation, is aimed at developing software to allow the elderly to communicate with robots that can respond to a wide range of verbal language, non-verbal gestures, and touch.

"If we can help the elderly remain independent and continue living in their own homes, that will improve their health outlook while relieving the burden on family members and health care providers," said Zefran, the lead researcher.

The researchers say that their communication interface software will have at its core a novel adaptive and reliable recognition methodology called Recognition by Indexing and Sequencing (RISq), which will allow the robot to comprehend speech altered by impairments and to learn and adapt to such speech.

To enable a robot to understand and correctly respond to various forms of human touch, the researchers will combine techniques from natural language processing and haptics, a scientific term to describe the computerized sense of touch.

They say that the robot will also know how to respond to the user safely when performing everyday chores, such as cooking or making a bed.

"We'll start by observing interaction between human helpers and the elderly. We'll identify what kind of language, physical interactions and non-verbal interactions are used. Then we'll develop a mathematical framework to model this interaction so it can be treated by the robot as a single way of communicating," Zefran said.

The researchers say that they will program and test a robot, in order to devise refinements, as the project progresses.

"The human-robot interface is really a long-standing, open problem that won't be solved in three years. But we'll have a working prototype by then, and we'll know what additional research needs to be done," Zefran said.

He believes that this research project may also find widespread use in delivery of institutionally based health care, where routine tasks now done by nurses could be handled by robots.

"If robots can alleviate some of the burden nurses face, they then could spend more time where they're really needed - providing the human contact that a robot can't replace," he said.

Zefran has revealed that his work will include developing seminars or a new graduate or upper-level undergraduate course that considers the various factors that allow robots to perform more sophisticated tasks.