Robot Arm Controlled by Neural Net

This is an experiment in which I'm trying to teach an artificial neural network about a virtual organisms' physiology. The learning process is guided by visual feedback: specifically 3d coordinates of a marker mounted on the robot.

I'm still working on getting the visual feedback in a usable way. The teaching process will be based on an evolutionary algorithm.

Category:

Howto & Style

Tags:

• Robot
• Arm
• Lego
• NXT
• ANN
• Artificial
• Neural
• Network
• Evolutionary
• programming

[karistajaify]

some1 can say me how i can buila ar robot who have neural net and laptop wher he save his mind i'am a biginner but i wanna make robot

[PsionNinja]

....Open Claw.....Insert Neck ;)

sweet setup

[Joe22c]

A neural net can does learn if there is an outside program that "guides" the neural net. Let's say our neural net is the computer program. A human could be the one "guiding" this learning.

Essentially, we could start off with a completely arbitrary and randomized set of "weights" for each neuronal connection, and over time, if we "correct" the machine for every desirable outcome and vice versa, the machine essentially learns, over time, by trial and error.

[Joe22c]

Information processing (IP) as a task is instead distributed every IP unit (modeled after neurons).

The unit, like the neuron, is essentially a switch. it's either "active" or "not" and whether or not it's active is a function of the amount of 'stimulation' (and inhibition) it recieves from its connections, as well as a set threshold level.

Thus the 'conclusions' a neural net reaches is actually a function of the activation patterns of the entire neural net.

[Joe22c]

The problem with the PSS is that it follows a strict set of rules. It doesn't really learn. (Actually, the *real* problem with a PSS is that it can't combine top-up and bottom-up information together, so while it can beat the world's greatest chess champion, it has a hard time recognizing a pattern that a child could recognize.

The next model of cognition is the Artifical Neural Network, or Neural Net.

This model is based on the structure of the brain. Essentially, there is no CPU.

[Joe22c]

A neural network is essentially one of two ways of modeling the mind, or consciousness.

The traditional method, physical symbol systems, was first proposed. It works essentially the way your computer works; it has a specific set of rules; a large random access memory (RAM) and a small CPU that takes informatino from the RAM, processes it, and puts it back. The power of a physical symbol system is that it does one task at a time, but it does this very quickly.

[vacuum20]

what soft did u use ?

[minousoft]

Awesome, how complex is the neural network?