Page 1 of 2

Brain kit

Posted: Wed Jul 31, 2002 12:23 pm
by hlreed
If you could buy a kit to make many brain structures for your robots, would you and how much would you pay?

Re: Brain kit

Posted: Wed Jul 31, 2002 2:54 pm
by bschwabe
I think you might be better off hiring someone to operate the remote control as far as finance is concerned.

Re: Brain kit

Posted: Thu Aug 01, 2002 8:30 am
by hlreed
You would rather hire an operator than building one. I will mark that down. Or, what did you mean?

Re: Brain kit

Posted: Thu Aug 01, 2002 10:09 am
by bschwabe
This was intended as a joke...<p>What I meant was... Depending on the cognitive depth in which you decide to "build a brain" you might consider "renting" a real, already working brain for the purpose. i.e. the guy with the remote control.<p>...although if you could "build" a working brain, money would probably not be an issue. ...because fortunately for you, you would be a person in very high demand. :o )

Re: Brain kit

Posted: Thu Aug 01, 2002 12:09 pm
by josmith
Do you have money in your world?

Re: Brain kit

Posted: Thu Aug 01, 2002 12:16 pm
by hlreed
Beau,
I can take a joke. However, I started working toward understanding how to build brains in 1958. (Marvin Minsky). It was in 1994 when I started over, looking for the least possible brain. (Brains are of different sizes, so there has to be a least one.) The least brain is simple comparison of two inputs. In fact that defines the requirements for a brain, having more that one input. I made an algebra for some possible brain functions. I made hardware to match the algebra. This is how you build brains, from the bottom up.
I am trying to survey the group to find interest, whether I should go now or have it wait another 100 years.
You might have some ideas on how to make nodes that would retail for $1.00 Think about it and I will forgive the joke.

Re: Brain kit

Posted: Thu Aug 01, 2002 12:20 pm
by hlreed
josmith,
I live in the same world. No I have no money.

Re: Brain kit

Posted: Thu Aug 01, 2002 12:45 pm
by bschwabe
If you were to design a structure in silicon with many transistors functioning as a single "neuron" whose transistor WELL was influenced by the outputs of the other transistors sitting within the same WELL either resistively or capacitively, you could theoretically have a structure capable
of "learning" based on the "popular opinion" of the other transistors. The device would in fact be persuaded to be an NMOS or PMOS device based on the inputs you have connected to the structure. Cascading and looping techniques could then be applied for other desired retention properties.

Re: Brain kit

Posted: Fri Aug 02, 2002 8:24 am
by hlreed
Beau,
The hooker is in the words you have in quotes, plus the word neuron.
Brains mediate between sensors and motors, or brain matches sensors to motor. There are more sensors than motors, so brains are functions.
Neurons are also functions: axon = f(synapses)
Note synapses is plural. This is what my nodes do. They combine into trees which can match any wild neuron. Some functions are: -, + , MAX, MIN,
and things like IS which compares n inputs and produces an output only if the inputs are equal.
Thanks for the input. Now, would you buy a brain kit, and how much would you pay?

Re: Brain kit

Posted: Mon Aug 05, 2002 4:23 pm
by josmith
What exactly would the kit do?(please answer in terms that we mortals will understand)

Re: Brain kit

Posted: Tue Aug 06, 2002 8:23 am
by hlreed
josmith,
The basic kit would provide the structure, like lego blocks, to learn how to build neuron trees instead of command control. You need at least one motor, one motor controller, one CNode, one battery node, two ADC's, two light sensors, 3 telephone cables for analog, 3 cables for data streams. From this you learn how to let your robot determine direction of largest or smallest light intensity.
From there you can learn to make your own motors and sensors run from this. You can add nodes now at any time to build any function you can imagine.
With the kit will be a booklet explaining Halalgebra and some experiments.
There are about 16 different nodes to experiment with.
Also you get me to help you.

Re: Brain kit

Posted: Tue Aug 13, 2002 7:18 am
by keith
For a simple neural network, you can use your 286 computer, or even with a basic stamp.<p>[ August 13, 2002: Message edited by: Keith ]</p>

Re: Brain kit

Posted: Tue Aug 13, 2002 12:30 pm
by hlreed
Keith,
Thanks for responding. I am trying to run a survey and everybody is trying to tell me what to do. What I have done is take programming out of the computer, into the real world. You program, which you can write on a napkin, translates into instructions to assemble the hardware from stock. The language is Hal algebra. The kits are the hardware that maps to the algebra. It just so happens that both map to neurons.

Re: Brain kit

Posted: Wed Aug 14, 2002 12:24 pm
by keith
Hi,<p>Sorry about my last posting. I just visited your site, and it sounds very interesting. But you already put price there, then why survey for "price"? Anyway, I liked your idea very much. If I can see photos or some demo of a robot on the site, it would be great. <p>Your idea and kit sounds good, but I can't really imagine how it would work and what it can do. Could you put some sample on your site, like a robot you built with the kit and what it does, etc.?<p>Thx,<p>KK<p>[ August 14, 2002: Message edited by: Keith ]</p>

Re: Brain kit

Posted: Wed Aug 14, 2002 12:43 pm
by hlreed
Keith
I have a ton of papers and such I could send you. First one would be Hal algebra so you can see the language. Give me your email and I will send you one. These machines do exactly what the algebra says. You say Doing = Do - Don't ; which is one CNode and feed anything into Do and Don't and Doing responds exactly as motor or servo.
My website has many places to comment. Please use it.
Thanks,