Communications of the ACM
Self-Taught Robot Figures Out What It Looks Like, What It Can Do
Researchers at Columbia University have developed a robot that learns what it looks like without any outside help.
Credit: Robert Kwiatkowski/Columbia Engineering
Columbia University researchers have developed a robot that learns what it looks like without any outside help.
The robot is equipped with a jointed robotic arm and a grasping hand similar to those used in many factories; the researchers deployed the robot with no previous training, and it behaved like an infant, moving randomly while attempting different tasks.
Within about a day of deep learning, the robot developed an internal picture of its structure and abilities. After 35 hours, the robot could grasp objects from specific locations and drop them into a receptacle with 100% accuracy.
Said Columbia's Hod Lipson, "If we want robots to become independent, to adapt quickly to scenarios unforeseen by their creators, then it's essential that they learn to simulate themselves."
From New Scientist
View Full Article
Abstracts Copyright © 2019 SmithBucklin, Washington, DC, USA
No entries found