Communications of the ACM
Ask the Crowd: Robots Learn Faster, Better With Online Helpers
Researchers at the University of Washington have determined that crowdsourcing is an effective way to teach robots.
Credit: University of Washington
University of Washington (UW) computer scientists have discovered crowdsourcing is an effective way to teach robots, suggesting robots could one day simply ask the online community for instructions to complete tasks.
"We're trying to create a method for a robot to seek help from the whole world when it's puzzled by something," says Rajesh Rao, director of UW's Center for Sensorimotor Neural Engineering. "This is a way to go beyond just one-on-one interaction between a human and a robot by also learning from other humans around the world."
The team's robots use machine-learning techniques to create models that require a significant volume of data, which can be provided via crowdsourcing. In one study, the team asked participants to build a simple model using blocks, then asked the robot to create a similar model. The robot could not complete the model based on the few examples provided, so the researchers turned to the Amazon Mechanical Turk crowdsourcing site. Using more than 100 crowd-generated models of each shape, the robots were able to build the best models of each participant's shape, based on difficulty to construct, similarity to the original, and the online community's ratings.
The researchers are now studying crowdsourcing and community-sourcing as ways to teach robots more complex tasks such as retrieving objects in a multistory building.
From UW News (WA)
View Full Article
Abstracts Copyright © 2014 Information Inc., Bethesda, Maryland, USA
No entries found