MIT Wants Robot Servants to Flexibly Juggle Chores
Rosie the robot maid may not remain a 'Jetsons' fantasy for too much longer. MIT computer scientists have honed a decision-making process that may help robots juggle diverse chores such as preparing dinner or loading laundry into the washing machine.
That means getting robots to do advance planning to accomplish their goals, but not planning out each step in such detail as to leave robots without room for flexibility. It's similar to how humans know how to get to the airport early and check in to board a plane, but don't plan their exact walking routes through the airport.
"We’re introducing a hierarchy and being aggressive about breaking things up into manageable chunks," said Tomas Lozano-Pérez, co-director of MIT’s Center for Robotics. [Read More: Top 7 Useful Robots You Can Buy Right Now]
The MIT approach creates a rough timeline of what robots may need to do, but plans detailed moves for only the first few steps. That may not be as efficient as a robot that follows a set of detailed commands like a choreographed ballet to tackle the household chores, but it allows for the unknowns.
"In computer science, the trade-offs are everything," said Leslie Kaelbling, an MIT computer scientist."What we try to find is some kind of 'sweet spot' … where we’re trading efficiency of the actions in the world for computational efficiency."
The idea may seem a bit counterintuitive to detail-oriented robotics researchers. But leaving certain details until later can help robots' limited brains deal with the tasks at hand, according to Stephen LaValle, a computer scientist at the University of Illinois at Urbana-Champaign.
"Often in robotics, we have a tendency to be very analytical and engineering-oriented — to want to specify every detail in advance and make sure everything is going to work out and be accounted for," LaValle said. “[The researchers] take a more optimistic approach that we can figure out certain details later on in the pipeline."
Next up, the MIT team wants to create learning algorithms that allow robots to figure out what steps to put off and which steps to prioritize. That could help household robots decide when to turn down the heat on a boiling pot and go answer a ringing doorbell – or it could help military robots make snap decisions on a rapidly changing battlefield.
"So it’s not strictly about getting a robot to do stuff in your kitchen," Kaelbling said. "Although that’s the example we like to think about — because everybody would be able to appreciate that."