Engineers at Columbia University in New York claim that an artificial arm with a grasping hand is a step towards self-awareness.

At the beginning of the study, the robot did not know if it was a spider, a snake or an arm. Robert Kwiatkowski explains

After around a day of "intensive computing", it built up an internal picture of its structure and abilities - a "self-simulation".
After less than 35 hours of training, this "self-model" helped the robot grasp objects at specific locations and drop them into a receptacle with 100 percent accuracy. Prof. Hod Lipson calls it the holy grail.

Even when relying entirely on its own internal self-model – its "imagination" - the robot was able to complete the pick-and-place task with a 44 percent success rate.

Other tasks included writing text with a marker.

Subscribe to our YouTube channel: https://goo.gl/jHqzVd
Subscribe to our Twitter page: https://goo.gl/Stnu74