In a major scientific breakthrough, scientists have developed a self-aware robot that is capable of operating without human instruction, even possessing the ability to initiate self-repair.
The robot was constructed by engineers at the Creative Machines Lab at Columbia University. The result of their research was published in the journal Science Robotics.
Previously, no robot it has been able to create an image of itself. But now that has changed. The new robot developed by engineers used a deep-learning algorithm to learn its identity over the course of a day and a half of intense computing. At the end of the 35 hours, the robot produced a self-image that was accurate within an inch and a half.
In addition, the robot learned how to do all of this without possessing any prior knowledge of physics, geometry or motor dynamics. Instead, the robot learned how to adapt. It used self-simulation to quickly adapt to tasks and a variety of things in its environment. It also used self-imaging to initiate internal repairs.
“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,” said professor of mechanical engineering at Columbia Hod Lipson in a news release.
In tasking challenges with an input signal (i.e., instructions), the robot learned how to retrieve objects and put them in a receptacle, doing so with 100 percent accuracy.
Next, the engineers tested the same kind of tasking, however, this time the robot had to rely on its own internal self-imaging, without instruction. In this scenario, the robot achieved a 44 percent accuracy rate.
“That’s like trying to pick up a glass of water with your eyes closed, a process difficult even for humans,” said one of the engineers, PhD student Robert Kwiatkowski.
In another test of self-repair, researchers attached a deformed component to the robot’s segmented arm. The self-aware robot identified the flaw and formed an updated model of itself.
“While our robot’s ability to imagine itself is still crude compared to humans, we believe that this ability is on the path to machine self-awareness,” Lipson said.
The researchers also pointed out that there is some inherent danger in building self-aware machines.
“Self-awareness will lead to more resilient and adaptive systems, but also implies some loss of control,” the researchers wrote in a paper. “It’s a powerful technology, but it should be handled with care.”
