Robot Uses Google To Think and Learn

Osamu Hasegawa, associate professor at the Tokyo Institute of Technology (Editors Note: in this case the acronym for the institution will not be used), has developed a system that allows robots to look around their environment and do research on the Internet, enabling them to "think" how best to solve a problem.

Inside is a feature called self-replicating neural network, or SOINN (Self-Organizing Incremental Neural Network), it can think as humans do when taking on tasks that it has never done before. It can make educated guesses and decisions based on it's past experiences and knowledge.

"So far, robots, including industrial robots, have been able to do specific tasks quickly and accurately. But if their environment changes slightly, robots like that can't respond. This robot remembers only basic knowledge, and it can apply that knowledge to its immediate situation. If it doesn't know enough, it stops, and reacts by saying, "I can't do this because I don't know how." So, if you teach this robot just the things that it can't do, it incorporates those things as new knowledge, and it can solve the problem overall, by including that knowledge."

“Until now, robots only successfully produced certain tasks that were programmed. They could not adapt to their changing environment. In contrast to this robot, it knew only the basic knowledge, then it could develop,” said Hasegawa.






By using the cloud (Google, specifically in this case), SOINN can use the data of other robots, humans and computers to analyse its environment and perform tasks.  Once multiple versions of such systems are brought on-line, the processes will develop exponentially as the robots are able to teach each other via the cloud.  This will include using additional sensor data and algorithms that would not be part of the robots off-the-shelf programming.  
As Martin Ford has commented:

 Human workers need to be trained individually, and that is a very expensive, time-consuming and error-prone process. Machines are different: train just one and all the others acquire the knowledge. And as each machine improves, all the others benefit immediately.

Diginfo