There’s a robot that learned what it looks like all by itself. It has a jointed artificial arm and grasping hand similar to those used in many factories, but what makes this one different is that it went on a robotic journey of self-discovery.
To begin with the robot had no idea what shape it was and behaved like a infant, moving randomly while attempting various tasks.
Within about a day of intensive learning the robot built up an internal picture of its structure and abilities. After 35 hours, the robot could grasp objects from specific locations and drop them in a receptacle with 100 per cent accuracy.
To do this the robot was allowed to readjust its movements as it went along, but even when that ability was taken away it was still able to complete the pick-and-place task with a 44 per cent success rate. “That’s like trying to pick up a glass of water with your eyes closed, a process difficult even for humans,” says Robert Kwiatkowski, a PhD student at the University of Columbia, New York.
To test whether the robot could detect damage to itself, the scientists replaced part of its body with a deformed version. It was able to recognise the change and work around it, with little loss of performance. Other tasks included writing text on a board using a marker.
“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,” says Hod Lipson at the University of Columbia, New York, who led the study. However, the robot’s ability to imagine itself is still crude compared to humans, he says.