A robot programmed by researchers at Cornell University can anticipate human actions and assist with tasks such as opening a refrigerator door or pouring a drink.
The robot uses Microsoft Kinect and a database of three-dimensional videos to scan a room and identify what action is taking place. The robot then gathers data on how various objects in the room can be used, predicts different scenarios, and determines what to do based on what move it anticipates a human will make next.
Researchers in Cornell's Personal Robotics Lab say the robot does this by generating a "set of possible continuations into the future [and] chooses the most probable." An algorithm helps the robot anticipate what a person will do next.
The robot has 82 percent accuracy in making predictions one second into the future, 71 percent for three seconds, and 57 percent for 10 seconds.
The ultimate goal is to have the robot learn to predict human actions on its own. "The future would be to figure out how the robot plans its action," says Cornell professor Ashutosh Saxena. "Right now we are almost hard-coding the responses, but there should be a way for the robot to learn how to respond."
From CBS News
View Full Article
Abstracts Copyright © 2013 Information Inc., Bethesda, Maryland, USA
No entries found