Okay. Maybe this is OTT ;-)
Okay. Maybe this is OTT ;-)
by Modereso
Tue Apr 10, 2007 6:57 pm
Hi
Thought I'd share (well I think so anyway) wonderful news with the forum here. I was toying with the idea of building a decent bot eventually, but today something quite strange happened. I remember not so long ago, putting some unwanted hardware in the loft. I didn't think much of it at the time. Today I got the urge (after 4 years) to check it out.
To my surprise I found a 'Gigabyte GA586VX' motherboard/controller (I presume it's a gigabyte board - but an old one) with a P100, 16mb ram, PCI/USB enabled and room for a serial etc. At first I thought it would make a wonderful controller, but then 'over the top came to mind'
I suppose if you are working on a huge industrial bot then it could come in handy, but as a home humanoid it would fit a beast! Perhaps I could utilize PCI/Serial but it would be one hell of a project. I've seen some nice connectivity on the PCI front if you're looking for lots of I/O. Any ideas? Perhaps I can find some use of it?
Mode
Thought I'd share (well I think so anyway) wonderful news with the forum here. I was toying with the idea of building a decent bot eventually, but today something quite strange happened. I remember not so long ago, putting some unwanted hardware in the loft. I didn't think much of it at the time. Today I got the urge (after 4 years) to check it out.
To my surprise I found a 'Gigabyte GA586VX' motherboard/controller (I presume it's a gigabyte board - but an old one) with a P100, 16mb ram, PCI/USB enabled and room for a serial etc. At first I thought it would make a wonderful controller, but then 'over the top came to mind'
I suppose if you are working on a huge industrial bot then it could come in handy, but as a home humanoid it would fit a beast! Perhaps I could utilize PCI/Serial but it would be one hell of a project. I've seen some nice connectivity on the PCI front if you're looking for lots of I/O. Any ideas? Perhaps I can find some use of it?
Mode
by Humanoido
Thu Apr 12, 2007 7:36 am
Go for the beast. 
It's a fantastic find. Its all about physical size, so if it fits the bot, then great! Also, if the box has all the original software, then that's good too. I remember seeing some outstanding articles on constructing controller interfaces based on parallel and serial, etc. output for these computers. I'll bet if you check the web you'd be likely to stumble across some of these. However, you'll have to do all your own support if something goes wrong with the hardware or software. But I think you were figuring that in.
humanoido
It's a fantastic find. Its all about physical size, so if it fits the bot, then great! Also, if the box has all the original software, then that's good too. I remember seeing some outstanding articles on constructing controller interfaces based on parallel and serial, etc. output for these computers. I'll bet if you check the web you'd be likely to stumble across some of these. However, you'll have to do all your own support if something goes wrong with the hardware or software. But I think you were figuring that in.
humanoido
by Modereso
Thu Apr 12, 2007 1:16 pm
That's what I thought !
I do see many options and numerous potential with this board. But, it's a case of finding the right project/design for the job. I too can see some success on the P/S front, and no doubt, a month of research might find me some parts to consider. Depending on how I feel when the build process arrives, I might turn to some bigger and better heavy duty servo controllers. I'm very realistic, so in all honesty I'm not looking to build a 4ft high beast 'alone'.
I'll keep you updated, but the main point is, I have an intact platform (good start).
I do see many options and numerous potential with this board. But, it's a case of finding the right project/design for the job. I too can see some success on the P/S front, and no doubt, a month of research might find me some parts to consider. Depending on how I feel when the build process arrives, I might turn to some bigger and better heavy duty servo controllers. I'm very realistic, so in all honesty I'm not looking to build a 4ft high beast 'alone'.
I'll keep you updated, but the main point is, I have an intact platform (good start).
by Modereso
Fri Apr 13, 2007 11:36 am
Humanoido wrote:4-foot high, isn't that about the size of the multiple million dollar Honda Humanoid Robot Asimo? I'd say you're going to need some heavy duty servos and a team by your side.
humanoido
!
And I'd say, I would be, (at this present time) living in dreamland if I tried to design/match anything like that - hence the need to put 'beast' into some realistic proportion. ha ha
On a serious note, it's not really about height and power. Size shouldn't matter too much, as long as it can accommodate the board. The real problem is designing a system with enough power to carry that board. I'm not even sure as to whether I'm going with a biped/humanoid design, just yet. I guess the fun of typical AI/Robotics, is the many options you can explore during the design process. I've already thought of some cool idea's I could incorporate with the board, use of P/S and PCI being among those idea's. Writing/adapting software (from scratch) to utilize various I/O's/interrupts will prove hard, especially without an existing OS, so I've decided to bung a small Linux distribution onto a small H/D. Perhaps the P/S front needs some attention, but I'm not even thinking about that yet. I might have a fish around Freecycle and see if there are any useful parts, up for grabs.
My main goal is to get something working, at least
by Humanoido
Fri Apr 13, 2007 2:25 pm
Modereso wrote: Size shouldn't matter too much,
But size does matter.
I was looking at some taller robots, such as Humanoid Manus,
(oddly enough they named the jpg image, robosapien.)
http://www.robhatah.com/images/robosapien.jpg
http://www.robhatah.com/products.htm
and these all have some winning aspect due to their size.
Of course, not everyone wants to fight their humanoid in
RoboOnes.
Yes, you have some really outstanding ideas. Keep at it!
Keep us informed of both your plans and progress. This
project sounds like fun and very exciting too!
humanoido
by Humanoido
Fri Apr 13, 2007 4:00 pm
Possibly we need some people to finance our projects. If the more powerful servos cost $50 each, then a 24 servo humanoid needs $1200 just for servos! Not to mention the controller board and mechanics...
In asia, the governments are financing humanoid projects. This is one possible route. I can think of some pretty cool projects that could result by an infusion of more money!
I wish there was a humanoid robotics consortium, with money investors, and projects could be selected for finance, including the small ones by brilliant inspiring amateurs. Well, I'm not calling anyone an amateur, but those ideas in your post are certainly brilliant and inspiring.
humanoido
In asia, the governments are financing humanoid projects. This is one possible route. I can think of some pretty cool projects that could result by an infusion of more money!
I wish there was a humanoid robotics consortium, with money investors, and projects could be selected for finance, including the small ones by brilliant inspiring amateurs. Well, I'm not calling anyone an amateur, but those ideas in your post are certainly brilliant and inspiring.
humanoido
by Modereso
Fri Apr 13, 2007 4:52 pm
No, no, thanks for the support!
I'm pretty much an amateur, unless I eventually go onto bigger things, of course. I do feel more at home exploring within the deep end, though, and do have plenty of ideas, which I think with some hard work could materialize. But like you have just suggested, the biggest problem will always be money. I think funding/backing is a good idea, but there would have to be some 'solid' foundation in place to prove a winning/leading research program. It's no good asking for funding just to develop a 'prototype', unfortunately, most organizations only fund ongoing research groups from lucky top-end projects in which they see potential. I wish that could change. But we can certainly learn a lot from aspiring ideas, though....definitely.
Another problem (as someone has already said) is that advanced AI/ANN is still in early stages. Most high-end developers / scientists are recognizing the use of ANN within larger projects, but they are still in very early stages. Sadly, whilst the movies are cool, many of them make the development/motion of advanced robotics look easy when actually, they take A LOT of man power to get right (as you know). Therefore, I like to ask the question, how far can we take homemade robotics?
Well actually, I think on a smaller scale you can do the job quite well. For instance, we have access to some pretty neat components, which we can control using custom made controllers/software design. And we also have the ability to develop a system for a reasonable price (compared to the Honda asimo, naturally). So in all honesty, a bread and butter project can grow into a much bigger thing. I admire anyone in the same boat, really
I'm pretty much an amateur, unless I eventually go onto bigger things, of course. I do feel more at home exploring within the deep end, though, and do have plenty of ideas, which I think with some hard work could materialize. But like you have just suggested, the biggest problem will always be money. I think funding/backing is a good idea, but there would have to be some 'solid' foundation in place to prove a winning/leading research program. It's no good asking for funding just to develop a 'prototype', unfortunately, most organizations only fund ongoing research groups from lucky top-end projects in which they see potential. I wish that could change. But we can certainly learn a lot from aspiring ideas, though....definitely.
Another problem (as someone has already said) is that advanced AI/ANN is still in early stages. Most high-end developers / scientists are recognizing the use of ANN within larger projects, but they are still in very early stages. Sadly, whilst the movies are cool, many of them make the development/motion of advanced robotics look easy when actually, they take A LOT of man power to get right (as you know). Therefore, I like to ask the question, how far can we take homemade robotics?
Well actually, I think on a smaller scale you can do the job quite well. For instance, we have access to some pretty neat components, which we can control using custom made controllers/software design. And we also have the ability to develop a system for a reasonable price (compared to the Honda asimo, naturally). So in all honesty, a bread and butter project can grow into a much bigger thing. I admire anyone in the same boat, really
by Modereso
Mon Apr 16, 2007 12:17 pm
Humanoido wrote:It's true, that we can do practically everything that the big labs do, only on a smaller scale. That's probably why we can accomplish so much with our smaller humanoids.
humanoido
I think typical kits / packages might be slightly on the edge of limitation, but this is where customizing/personal design comes in. Build a basic platform and then upgrade it. I've seen some nice modifications to the RN-1 for instance, and a similar bot with target/speech/recognition and
coordination. And as for ANN, it's pretty self explanatory what software modeling can accomplish.
Even if the bot spends most of his time wired up to an ANN backbone, it's still impressive to see what he does. I think I could spend hours of fascination with the rn / khr / Bioloid alone lol
by Humanoido
Mon Apr 16, 2007 6:55 pm
Yes, you did mention three of the top selling humanoids. It seems that many companies are making a humanoid product just to get the platform out there. So we're seeing a lot of different approaches and standards, all going for the same market. It's a lot like early computers, when everyone seemed to have a product.
I am seeing more and more applications which are mainly on computer internet, and there's a trend to put software there rather than on the main computer. You can even edit your photos now online. Perhaps this is a good approach for robots too. I mean you can get control, processing, speech, intelligence and a lot of things from the net. What do you think?
humanoido
I am seeing more and more applications which are mainly on computer internet, and there's a trend to put software there rather than on the main computer. You can even edit your photos now online. Perhaps this is a good approach for robots too. I mean you can get control, processing, speech, intelligence and a lot of things from the net. What do you think?
humanoido