Communications of the ACM
Robot 101: Learning to Work With Humans
"We're looking to develop fast, smart tasking algorithms so robots can work interdependently with people," says MIT's Julie Shah.
Credit: Technology Review
Massachusetts Institute of Technology Interactive Robotics Group director Julie Shah is focused on teaching inherently safe robots to work in teams with people, and vice versa. "We're looking to develop fast, smart-tasking algorithms so robots can work interdependently with people," she says.
Shah envisions future factories where industrial robots work safely with humans by putting them on mobile bases and rails thanks to new safety standards and technology. She points out today's cutting-edge robot training methods rely on demonstration and interactive rewards, but her research indicates robots are frequently unclear as to what the reward is referring.
Shah studied training techniques of human teams to improve those for robot training, and learned cross training was a consistent factor in effectiveness. Her research team tweaked modified learning methods and algorithms so the robot gets input by switching roles with the person, rather than receiving it as a positive and negative reward.
Shah says the experiment yielded improvements in objective measures of team performance, gains in concurrent motion between human and robot, reductions in idle time, and subjective improvements. Shah also aims to optimize task planning and deployment in hybrid human-robot teams through her research.
From MIT News
View Full Article
Abstracts Copyright © 2014 Information Inc., Bethesda, Maryland, USA
No entries found