Blog

News


Zurück zur Übersicht

30.05.2019

How does one create and implement an ANS? Part 1

An ANS - Artificial Nervous System, is similar to
a BNS - Biological Nervous System, therefore
a comparison between a digital implementation and its
corresponding analog counterpart is very useful.

First of all, the analog one is always a template for the digital one,
Why? Because it already exists in nature, in various forms and bodies.
Yes, bodies, they are the home of the BNS, cause without a body, no BNS.
So, BNS is always part of a certain body, it can NOT exist without it.

Now, the BNS consists of 3 main parts: brain, sensors and actors, means
Get information from the environment via the corresponding sensors,
then process the information in the brain and find out a response,
which will be transmitted to the actors, in order to perform a reaction.

So, without any sensor, there is no information from the outside world.
Without information, there is no need to perform any processing, right?
This also implies that a processing is only active, when information is provided.
Without an actor, there is no response to the performed processing.

Now, in an ideal world, an ANS should be the digital implementation of a BNS.
Question? Which BNS should we take as a template? Maybe the human one? he he.
Unfortunately, the actual status in science today, does NOT provide any explanation
on how a hunan BNS is processing information, let alone the internal representation.

And there is another thing, namely even if a human BNS is quite complex,
it was not created in one step, but it was developed in time during evolution.
Evolution? Yes, from a single cell via multicell organisms, to more complex ones,
like prehistoric creatures, and after that fish, mammal, primates and finally humans.

But the development of the BNS is NOT only the development of the brain itself,
instead, it is the development of the entire body, with the connected senses
muscles and adjacent limbs, in order to complete the BNS and connect it
to the corresponding body, with different capabilities and features.

Now, different stages of evolution create new organisms, what we call species.
If one wants to create a human BNS, it also needs to understand its evolution.
And I mean not only the evolution during the hundreds of millions of years,
but also its evolution during its own lifetime, from inception till expiration.

Now, a human life can be various, means different stages of knowledge
and adequate intelligence can be achieved, usually by going to school,
first primary then elementary, high-school and even a degree, maybe even PhD.
Not to forget the different areas of specialty and the performing jobs.

Now, let us take a look at the implementation of an ANS. What do we need? He he.
First we need a body, then we need the corresponding BNS, means sensors, brain and actors.
And we need a specialty, which will be needed, in order to perform a specific "job".
And the required knowledge, which has to be accumulated via (self-)learning.

OK, lets go backwards. We want an automated vehicle, self-driving. So?
We need a body, which is the car itself, or better an electric car.
Then we need the sensors and the brain. The actors are already there.
Means the electric motor, the rechargeable batteries and the wheels.

Now let us analyze the requirements of the needed brain and its knowledge.
Which BNS can we take and use it as a template? Maybe a mammal? That is oK.
A mammal does have all the required functional structures needed for.
A horse BNS would fit in great. Only we need to adjust it, to fit in the requirements.

As an analogy, we can take a whale BNS and fit it in to be used as a ship ANS.
Same we do with a bird, to use it as a template for autonomous planes, right?
But wait? What do all these vehicles have in common? Right, they do not need to talk.
And they would replace our existing human drivers and pilots, to enable autonomy.

OK, what about human alike robots. Can we do it also? Sure. Why not?
Take a human BNS as a template, adapt it, and create a robotic worker.
OK, we are done, right? Not quite. Why is that? Because there is a problem.
Problem? Yes, we human have a big problem. We have no idea how a brain is working.

Oh, really? But humanity has achieved so much. We went on the moon.
We discovered the DNA, we split the atom, we create cities and civilizations..
Yes, that is correct. However, regarding machine made intelligence
we have no idea, cause our masterpiece is a conventional computer.

How does a conventional computer work? well, one instruction after the other.
The same like our consciousness works. So we took the human BNS as a template.
Yes, but we forgot something important, namely the consciousness needs
the subsonscious, which does 99,99 % of the brain work, he he.

Ok, we know how the subconscious works, namely massively parallel,
while the consciusness works only serial, means one after the other.
OK, so we need massive parallel execution, instead of single parallel.
How do we do that? Which instruction set is to be used? Good question.

If we take the BNS as a template, then we discover that is is actually,
a network of many networks, each of them consisting of biological neurons.
OK, we can do digital neurons, means as digital implemention of template.
Simple, right? Not quite. Why not? We know how they work, each of them.

Unfortunately that is not enough. Why? Cause we have no idea, of
how the neurons build networks, or even small groups of neurons.
Not to forget the fact that we have no reasonable explanation of
how these networks are creating our most precious "consciousness".

Lets assume, we have solved our big problems, how do we create an ANS?
First, we have to create something, which executes neural nets in parallel.
Today, the only possibility to do this is the FPGA, means programmable chip,
while the corresponding development language is called VHDL, he he.

Second, we need the sensors and actors, to be connected to the brain.
OK, we take digital cameras and digital servos and some framework for the body.
Afterwards, we connect all these things to what we call an embedded system,
means place the brain inside the body and add a rechargeable battery.

Next time, I will talk about the development of the brain of the ANS.



Zurück zur Übersicht