The idea of robots folding laundry or cleaning dishes has probably been popular since Rosey made her debut on “The Jetsons” in the ’60s. But making it a reality is much more complicated than some might think.
A professor at Arizona State University is trying to figure out how to program robots to do the tasks humans don’t want to do.
Sid Srivastava teaches computer science and engineering at ASU. He said as humans, it’s easy to see someone move one cup from the counter to the sink and understand how to clear the whole counter, but:
“This is a very hard generalization task for robots because I’m showing the demonstration in one cup and one situation in one location, and I’m expecting the robot to generalize that to clear the entire counter without any additional instruction,” Srivastava said.
He and his team are trying to figure out how to program robots to generalize those concepts and be able to repeat them in different locations because every home is different and each kitchen has unique measurements. Each new location would usually require a highly skilled person to program the robot for those specific conditions.
“That’s what makes it difficult to deploy robots in scenarios that are less controlled,” Srivastava said. “You can deploy them in factories because factories are controlled environments.”
He and his students are developing new methods to solve that problem. The research is funded by multiple grants from the National Science Foundation.
