Saying goodbye.

@robotbuilder

Goodbye Casey, and thanks for all the fish 🙂 I hope you find what you're looking for.

@robotbuilder Goodbye, I will miss the humour.

Hi Casey @robotbuilder,

   I can understand your frustration, but please don't give up yet.

Obviously, I have no influence or idea, where Bill or the other forum contributors are going to lead in the future. I haven't yet worked on AI, but in terms of robots, whilst for many years there have been professional research groups, commercial and academic, with healthy budgets and teams of contributors,  yet the commercial 'mass-market' products have largely been things like vacuum cleaners and lawn mowers .. useful no doubt, but not very exhilarating. Industries, including farming, also have their 'grown-up' versions, like automated tractors, but they too have a fairly simple brief.

Self-driving vehicles have shown remarkable progress, though they have some unfinished work. Whilst they are presently feasible in controlled environments, dealing with the undisciplined road systems and users is a very challenging goal. Of course, we expect these machines to be almost infallible, whilst human drivers have thousands of accidents every day, so maybe our expectations are unrealistic.

So where does that leave amateurs with limited budgets, experience and resources. The presently successful, lowish cost, systems with simple goals, are able to operate with limited compute power, the others (presumably) have access to a considerable amount of 'computing horsepower' that precludes those with budgets that are below the billion dollar level.

Presently, various companies appear to be trying to bring 'horsepower' to smaller, cheaper chips, particularly for IoT applications. My crystal ball is too cloudy to see how far that will take us, but I remain optimistic. To some extent, I feel the limitation is finding applications which are more demanding than navigating a cleaner around a room, but not as demanding as driving a car down twisty country lanes and off-road. i.e. It is a technology looking for a purpose. 

The best of the LLMs (Large Language Models) are beginning to show some remarkable behaviour, whilst the not so smart ones are collecting the computing equivalent of egg on their faces. They too require vast servers etc. , but maybe the best efforts give a tiny view into how versatile AI may prove to be.

Can we make useful robots that do more than sweep a carpet? ..  It is probably difficult at present if we are relying on a processor board that costs about the same as a cup of coffee. I note that Jetsons seem to have become an expensive endangered species, but maybe a new species will emerge that has lot more 'horsepower' at reasonable cost. In particular, AI may benefit from a renaissance of analogue computer approaches, re-energised by semiconductor technology developed for digital computers, so that thousands or millions of analogue processors can be squeezed on to a square centimetre of silicon or similar. That could be a gamechanger!

And a large part of robots are about sensing their environment. A handful of ultrasonic sensors may help you reverse your car into a parking space, but that is like a one-off party trick. Certainly newer sensors are emerging, though how many are readily available in the "few dollar" bazaars in one-off quantities, I am less clear.

--------------

So if you decide to take a break from the forum, I will miss your contributions, but I wish you well. I hope you will keep in touch, and maybe propose an idea or two that is both feasible and interesting, as well as sharing anything else you think may be of interest.

But don't count success by the number of people that like your suggestion .. it is difficult to find a place that is not already filled with 'influencers' and 'populists' who can spin snake oil into wine, and collect millions of likes, but whose useful contribution is always 100% negative junk, sometimes with tragic consequences.

Really good ideas and suggestions often get derided by 'almost everyone' for years ...

Best wishes and good luck in whatever path you take, Dave

@davee 

I am not frustrated it is just a realization I am not on the same page as others here with regards to robots and AI and that is fine.  As I haven't kept up with the latest hardware offerings that everyone seems to be using I can't really make any useful contributions with regards to their use or the software required.

In my spare time I am again experimenting with my vision based robot so all is good.

Thanks again for your thoughts.

All the best, Casey

@robotbuilder It is indeed a wise man who knows when to withdraw. I think about doing that every week. Competing with the educated folks here is very hard being self-taught and a visual learner..

Hi Ron @zander & Casey @robotbuilder,

   It is not about competing, or needing to be on the same page. A little competitive spirit or agreement from peers sometimes provides a touch of adrenaline, but it is not essential. The diversity adds to the richness ... all those who bring a chink of their experience, and hopefully get some insight into what others are thinking or doing, is what makes it interesting. There are tasks that require the entire 'team' or 'group' to be acting as a single body, but those groups rarely produce anything innovative. You both have made, and I hope will continue to make many such positive contributions. Take pride in your achievements.

Best wishes, Dave

@davee Yes?

Hi Ron @zander,

  sorry, a badly timed keyboard click!.. message now added above

@davee Or perhaps Freudian?

I don't think there is a lack of interest... on the Internet or even here.  Looking at @thrandell topic, Genetic Evolution of a Neural Network Driven Robot has over 1700 viewings for only 6 pages of content.  I'd say there is a lot of interest, but it is likely a subject that people want to know more about, but are unwilling to dive in and actually work with it.  Many might think (or in my case... thought) it is some cryptic, difficult concept needing super computers with millions of parallel CPU's.  

I see many on the Internet discussing it, teaching the basics and in its most basic forms, the concepts are really quite easy and can even be run on any modern Mac/Windows machine up to a certain complexity level.  Heck... for a simple obstacle avoidance robot, it can actually be run on the MPU.  It really doesn't get more complex (programming wise) with more complex problems, it simply takes more time or the more raw horse-power that we've come to hear for things like ChatGPT, et al.  

Some people here confuse lack of forthcoming information with lack of knowledge or lack of interest.  I would suggest that is falls under the category of, "Those who can, do; those who can't, teach."  With the Google's of the world offering $900K/year salaries for AI engineering expertise... yes, there is one hell of a market.  Even the paper @thrandell topic uses as a reference, all the concepts are abstract and no written code is offered.  They're keeping it close to the vest even though, I would say their work is far less complicated than say a robovac. 

Maybe, @thrandell has some overall grand plan with his swarm robots that is quite marketable.  I don't blame him one-bit for keeping it close to the vest.  It would be severely tragic for him to do all the work and research, display it here and for some un-ethical visitor to this forum to take his work and put it into production.  I asked him where he saw his swarm robots going.  I probably crossed that proprietary line.  @robotbuilder asked him for his code.  Possibly crossing that line.  

For those same reasons, in my instance... I will offer up my source code for my version of the AI obstacle avoidance robot, InqEgg.  Considering I can write it with what I already know, I don't think it is beyond anyone here's ability to research/learn the techniques within say... a week or two.  That's all the time I have under my belt mixed with other life's duties.  But when I start getting into upgrading it with the house/building mapping type capability, I'm unlikely to do anything more than use abstract wording.  I simply see way too much of a market for such a bot and for my expertise (at that future point) for that capability.  

Back in the 1970's when I wrote my own Pong on an Apple II, I thought that such a trivial game couldn't be marketable.  Boy! was my engineering mentality wrong!!!  This is another Pong moment.  Any software capable youth looking for a future needs to get on the band-wagon pronto quick!  Even at my old-geezer age, I'm seeing opportunity even though, I'm not needing a paycheck.

@inq 

Even the paper @thrandell topic uses as a reference, all the concepts are abstract and no written code is offered.

I didn't know that.  I thought @thrandell was using code from a book.

@robotbuilder asked him for his code. Possibly crossing that line.

That never occurred to me.  I assumed he was just duplicating the code in the book. I have had an interest in AI most of my life and am able to draw on that knowledge to write my own AI including my own version of a GA program. It was interesting to watch the system iterate itself to a working set of weights.

As for anyone being interested in anything I write all I can say is without feedback you may as well be talking to your hand or to yourself so there is no point in it. 

Back in the 1970's when I wrote my own Pong on an Apple II, I thought that such a trivial game couldn't be marketable.

Although I never play computer games myself I found writing my own version of simple games (Pong was one of the first on the TRS80) was a good way to have fun learning to program. I didn't copy code just the behaviors with my own code and I drew my own graphics as I sort of like doing art. The hard part of writing a computer game is actually coming up with the idea and game play something I wasn't able to do.

Hi @robotbuilder,

RE: It was interesting to watch the system iterate itself to a working set of weights.

As for anyone being interested in anything I write all I can say is without feedback you may as well be talking to your hand or to yourself so there is no point in it. 

When someone, like yourself or @inq, does something original or out of the usual, then for me, there can be two or more levels that I may be interested.

With AI (for example), I am not currently working on it, so the minutiae of a long programme listing is unlikely to be that interesting, but I may be interested in the general scheme of how it works and what it achieves. I try to gradually build up a picture of what is possible and the underlying technology, even if I am not presently involved in it.

However, at some point in time, then I may be interested in some involvement .. at which point, access to software code, etc. will be a lot more valuable and interesting.

So apologies if I have not provided sufficient or appropriate feedback, but that is not neccesarily a reliable indicator of the value of your contribution. Please appreciate that although we live in world that can communicate anywhere in a fraction of a second, it may be days, weeks, months or years between somone authoring something and someone else appreciating the value of it.

Best wishes and take care, Dave

Posted by: @robotbuilder

↑

I didn't know that.  I thought @thrandell was using code from a book.

I don't remember reading that in his thread.  I just know the sited paper doesn't have any real meat in it.  I remember asking if he had an additional reference for that paper, but he didn't respond to that.  But, @thrandell commented Goldberg and the paper referenced Goldberg and it did use a lot of the same terminology.  That is why I got the book and it does have the engine code in Pascal.  That is back when I found the C++ version and posted that in his thread... or maybe in your other thread?... Anyway... somewhere.  

Posted by: @robotbuilder

↑

@robotbuilder asked him for his code. Possibly crossing that line.

That never occurred to me.  I assumed he was just duplicating the code in the book.

Obviously, I can't speak for him.  I'm sure he doesn't hang on our every word and might of missed those posts (yours and my requests).  But, I appreciate what he does share with us; and I also don't blame him one bit if he is working toward some marketable (or even self-education) end-goal.  There are way too many unscrupulous people that will gladly take his work and sell / palm-it-off as their own original work.  Again... practical / useful AI and the people that can write it are in astronomical demand.  

Posted by: @robotbuilder

↑

As for anyone being interested in anything I write all I can say is without feedback you may as well be talking to your hand or to yourself so there is no point in it.

Your position on your limits of helping/giving code is well known here and doesn't surprise me by being explicitly stated now.  

@davee 

With AI (for example), I am not currently working on it,

Like I wrote my initial interest here was in response to Bill's intention at the time to build a "real" robot which would have involved some AI and I hoped to glean useful information mainly on the hardware side.  I was looking for information to interface motors and sensors to the Arduino and RPi and his videos provided such information.

I see from your first post that "Science, including electronics, has been a major part of my life, and now I am retired I am enjoying the opportunity to look at things I didn't have time for in the past."

A glanced at some of your posts and I recall your input to my problem of getting an accurate read from the wheel encoders.