Scientist Rajesh Rao is trying to make a robot that can learn by trial and error, much like infants. If the goal is achieved, robots could adapt instead of having to be reprogrammed. Over two dozen labs around the world are now concentrating on this branch of robotics. Rajesh Rao's main goal is to make a robot that can "gaze track," or following another's gaze. He's trying to get a Japanese robot with cameras for eyes to identify what someone is looking at by following their eyes. The robot "tries" to identify the object, recognizes it, and says it out loud. Instead of Rao trying to improve the imperfect system through programming, the robot itself makes a "map" of what the person prefers to look at, which guides the machine, allowing it to learn. Rao also wants to see if a robot can learn to imitate other's actions. Scientists know that to make robots as human-like as possible and integrate them into society poses many challenges.
The experiment in this article is an example of hypothesis science, because even though the hypothesis wasn't stated, that was the general method taken. Based on observations with human infants, Rajesh Rao and other scientists hypothesize that they can apply this system of learning to robots. To try to prove their hypotheses, they conduct different experiments that may succeed or fail with different robots or technology.
The importance of this research to humans is that these robots, if perfected, can be made to better help humans such as the elderly, the sick, and children.
Author: Bruce Bower
Title of Article: Meet the growbots
Journal: ScienceNews
Date published: January 29, 2011
Pages: 18-21
1 comment:
Robots learning? But they don't have neurons, or brain matter, or feelings...Yes, they can probably use technology to 'decipher' (like the maps) and figure things out, but it's not like they're LEARNING. And integrating them into society would be more problematic than beneficial.
Post a Comment