COLLEGE PARK, Md. (CBS DC)– The concept of robots is fascinating to many and now researchers are trying to make them not only cool, but smart too.
Developers and researchers in the Autonomy Robotics Cognition, run by the Institute for Systems Research and the University of Maryland Institute for Advanced Computer Studies, are looking to advance the capabilities of robots.
University of Maryland explains that having a robot perform a single task is not the difficult part, but programming it to act on instinct is a broader challenge that is being explored.
“We’re trying to build the next generation of robots,” Yiannis Aloimonos, a computer science professor, told UMD. “These are robots that can interact with people naturally and do a variety of useful things.”
Aloimonos and his team are looking to revolutionize the way that robots function. Individual programming for separate tasks is currently needed, but researchers are hoping to develop awareness and common sense in robots so they are more useful in the day-to-day routines of human beings. The goal is for a robot to react to its environment naturally.
Researchers in the ARC had a robot cooking demonstration where the machine learned its culinary skills from observing dozens of cooking videos from YouTube.
They first attempted having the robot mimic the individuals from the videos, but this proved to be unproductive.
“We discovered what we needed to imitate was not the movement—it was the goals that existed within an activity that had to be imitated,” Aloimonos told UMD.
This led researchers to create a system where robots identified important objects and actions from the videos. The plan is for a robot to analyze how a person makes decisions, then string together the observed actions to form a completed task.
Don Perlis, a computer science professor who specializes in an area of artificial intelligence known as commonsense reasoning, is working with the team to make machines smarter.
“They were saying, ‘We need to add reasoning into our vision systems,’” Perlis told UMD, “and we said we need to add perception into our reasoning systems.”
The development of common sense in robots would open the door to machines dealing successfully with circumstances that are uncertain and ill-defined.
Systems have been triumphant in the area of performing a single task correctly, but artificial intelligence still lacks understanding.
“That view says the world is so complicated there’s not going to be a perfect set of axioms to get everything right,” Perlis says. “You may have figured out everything yesterday, but today something will be different.”
When robots are faced with uncertain or unexpected conditions, task completion can end up being a waste of time, if the robot is able to perform the task at all. This is an issue Perlis hopes can be fixed soon through giving robots flexibility.
“They’re not going to do everything perfectly, and humans don’t either,” he says. “But we’d like them to muddle through and get things done.”