Scientists have created a robot controlled by a biological brain made of rat neurons.The robot, named Gordon, is not exactly an Einstein but represents a remarkable bridging of the gap between biology and technology. Gordon relies a dish with about 60 electrodes to pick up electrical signals generated by the brain cells.

The brain drives the robot's movements.

Every time the robot nears an object, signals are directed to stimulate the brain by means of the electrodes, the researchers explained in a statement released today by the University of Reading in England. In response, the brain's output drives the robot's wheels left and right, so that it moves around in an attempt to avoid hitting objects.




...

"One of the fundamental questions that scientists are facing today is how we link the activity of individual neurons with the complex behaviors that we see in whole organisms," said Ben Whalley, a pharmacist at the university and member of the team that built Gordon. "This project gives us a really unique opportunity to look at something which may exhibit complex behaviors, but still remain closely tied to the activity of individual neurons. Hopefully we can use that to go some of the way to answer some of these very fundamental questions. "

http://news.yahoo.com/s/livescience/roboth...punsvgdvQYDW7oF

Wow, that's fascinating. Thanks for sharing.

So maybe Evangelion is not as far fetched as we might think... (computers running on human brains, robots given human personalities, etc.)

they did a similar project with a rat brain and a flight simulator in which they just hooked it up and watched the rat try and control the airplane. it apperantly did a moderately good job. (after ALOT of crashes)

it just kept on doing better and better.

The article is vague on many points.

First, how are the neuron cells kept alive - is there a blood based nutrient delivery system?

Second, how are the neurons connected? What is the neural network topology?

Third, how are the neurons stimulated in a reward/non-reward scenario (i.e. when the robot hits an obstacle)? Which ones are stimulated?

Fourth, is the assumption behind this that a network of biological neurons will learn in a way similar to a neural net?