Thanks, so sad to hear this news :-(

https://forum.dronebotworkshop.com/general/news-from-robopis-sister/paged/2/

Posted by: @robotbuilder

Sadly Robo Pi passed away some time ago.

Oh no, that's very sad to hear.
Last time I was here before my lengthy break, I thought he said he was doing ok with optimism for recovery.

My condolences to his family and everyone who knew him.

Sadly Robo Pi passed away some time ago.

With regards to your ambitions I would say you don't really appreciate the difficulty of the subject that is why after millions of dollars and input from some of the best minds real robots of the kind you allude to are very much in their infancy.

You might have seen James Bruton videos on building robots?
https://www.youtube.com/c/jamesbruton/featured

His attempt at a more practical robot that could actually do something useful.

https://www.hackster.io/news/james-bruton-s-really-useful-robot-in-progress-d4300b7b1bde

Not sure what if any advantage c# has over c++ a language I can already use.

Although an old thread, I've been studying it for hours and found it to be very informative.  As far as the title, "What is Artificial Intelligence", I would not want to weigh in as a combative, and certainly can't argue the points as well as the two main contributors.  For my practical needs, I will simply be using that AI for my robot will:

1. take in various stimuli 
2. try different things with stepper motors, servos, on its own
3. verify with sensors
4. grade itself on how close it comes to its pre-defined goal
5. rinse and repeat till it gets it right.

I can see a hardware/software approach to the above, simplistic algorithm.  But for the life of me, I can't see how you'll approach the problem getting your robot to ask you to help it.  I'm anxious to see how it comes along.

C# and C++

I can weigh in on the differences between (at least Microsoft's) C++ and C#.  True C++ (not their C++.NET) compiles to machine code.  C# compiles to what's called Intermediate Language.  It is almost human readable.  Therefore it has to compile a second time at runtime.  So it's somewhere between a compiled language and an interpreted language in speed.  Although I don't use Python, I understand it is an interpreted language.  Although, they may have created a compiler for it as well (I don't keep up with it anymore).  

For iterative design, Python is usually considered to be easier and faster to build the program.  But being interpreted, you may be talking several orders of magnitude slower when running.  This may not be an issue with things requiring interacting with us slow humans, but I would say the ANN, as it grows has exponentially more things to process and thus, you should use the fastest thing when running.  From slowest to fastest that would be Python, C#, C++, C, Assembler.  However, programming skill also plays a significant role.  It is possible that "bad" C code will be slower than a "good" C++ code.

RoboPi: To be perfectly honest with you at 70 years old it's simply too much for me to try to keep up with it all.

I don't even try to keep up with it all. I intend to work with what I already have. I would test my semantic a.i. ideas on my current computer.

RoboPi: Today there is so much technology available a new difficulty has arisen. Which ones to choose?

I am choosing none of them. I will try and do something with the technology I already have and understand.

There is nothing I can do with the Arduino and RPi that I could not already do with a laptop and the K8055 plus some extra circuits made out of discrete ICs like I used to use on the old DOS machines using their parallel ports.  I am however using the Arduino Mega because I have one and it is easy to use and program.  I also have the RPi 3 and again have learnt how to use the gpio pins on that.  I think that is all I will need at this stage to make a working robot.

Of course dreaming is a pleasant pass time and if spending your time and money buying and learning about the latest hardware is sufficient then be happy with that.

If I am successful enough with the robot base I will then give it a nice friendly looking robot shell with the flashing lights :)

The cool looking ones with pretty flashing lights

That's what I'm thinking too.  If I ever manufacture hardware ANNs I'll be sure to include pretty flashing lights.  Otherwise they won't sell. ? 

Which ones to choose?

That's easy

The cool looking ones with pretty flashing lights

Extra points if they fly too

I already have Jupyter Notebook and Anaconda Python on my computers along with the entire science packages of Numpy, MatPlotLib, and Pandas, etc.

I don't have CircuitPython, but as far as I can see it's designed for special boards, specifically those being developed or supported by Adafruit.    It's also an attempt to bring low-level hardware programming into the reach of people who don't understand how the hardware works.   If you understand how the hardware works you can use existing Python language to program it.

It is amazing though where all this is going.  Adafruit (and many other companies) are creating more powerful boards all the time and creating them with smaller and smaller footprints.   This trend isn't likely to slow down anytime soon.

To be perfectly honest with you at 70 years old it's simply too much for me to try to keep up with it all.   I try to pay attention to what's going on just to have an idea of what's available in case I see something that would fit into my projects seamlessly.   But what I'm finding is that I'm probably better off just designing my own specialty boards using KiCad and smd components.   This way, not only will the boards be perfectly tailored to my projects, but I'll also have a full understanding of their hardware requirements since I will have designed them from scratch. 

One thing that is kind of interesting though is that if I wanted to manufacture my boards for sale to the general public it would be wise to design them to be compatible with something like CircuitPython to make them more attractive to consumers.  So that might be a good reason to learn about CircuitPython anyway.

But as I say, my brain isn't a 20 year old puppy anymore.  So I need to put practical limitations one exactly how much I try to learn about.

I'm basically getting into this too little too late.  But that's what had to be.  Back when I was a 20 year old puppy these technologies simply didn't yet exist.  Back then we were still using large discrete components and crude programming languages.   Usually assembly language if you wanted to do anything on the machine level.

Today there is so much technology available a new difficulty has arisen.  Which ones to choose?