Slashdot Mirror
VIA-based Mobile Robot Design For Download
An anonymous reader writes "This article at DeviceForge introduces the "PC-bot" -- a new mobile robotics demonstration platform design that is being made available for free download from Roboteq. PC-bot was designed to be highly flexible and extensible so that it can serve as the basis for a wide range of customized mobile robot designs featuring web cameras, motion detection software, and more. Details on the mechanical, electronic, and software design are available at Roboteq's website. Currently the bot runs Windows, but Roboteq expects that one of the first enhancements from the user community will be Linux support." Unfortunately, there's no way to download the RoboteQ motor controller the design requires -- that will cost you about $500.
So does that mean it'll be buggy and incompatible upon release and require monthly 4n1 updates to get working properly by the time it's two revolutions behind the tech curve? :D
will the 'bot be sentient enough to turn off it's power for good? :)
That's CHEAP
Then again, you could probably get away with some Lego Mindstorm cores and do the same thing.
Asimo, it ain't.
Intel and AMD may have the fastest processors, but I'm much more interrested in Via's small and cool (low temperature) processors... The possibilities just seem to grow with each generation :)
The new Via Eden-N is especially interesting.
.: Max Romantschuk
hmmmmmm a robot running windows...
why doesn't that sound like such a great idea?
oh yeah! that whole fear of becoming borg!
you call him Longhorn, i'll call him Tex.
So, does this robot also transform into a jet fighter?
YLFIOne god, one market, one truth, one consumer.
She ended up being spyware. I'll admit she was sexy though.
how would it fair over here?
I see no flame throwers or hydraulic crushing devices.
I don't want to start a holy war here, but what is the deal with you VIA fanatics? Masturbating with a hand full of broken glass for about 20 minutes now while my mother attempts to load a 17in pizza in the oven for 20 minutes
Interestingly enough, at the robotics lab at Utah State University (http://www.usu.edu/) we were building general designs like this, and then another team was taking those and designing software to make the robot sniff out car-bombs. The robot would scan the undercarriage of the car and use some cool sensors.
Then again, I've had issues with VIA myself . . . sometimes their chipsets do weird things. Unnatural things.
Roboteq expects that one of the first enhancements from the user community will be Linux support.
so Roboteq wants linux users to purchase it for $495, and write their own code for it? Sounds awefully cocky to me. "Buy it, write the software for it, pass it on to us, and we'll distribute it to the masses, and cut $100 or more from our expenses per unit. We'll then market it to the Linux masses, and we may or may not pass that price break on to them, as well."
is it wishful thinking to hope that these companies might donate a few of those puppies to a major, or even a few minor sized open source shops and let them write the software for it and get some well earned name recognition?
I downloaded "RoboteQ.molecular.blueprint[45A99B28].bz" last night, and my Zyvex FX3000 nanoassembler burn^H^H^H^Hbuilt it in only 6 minutes! Sure, it's still technically illegal to copy someones "Intellectual Property" (until HR837475 becomes law), but putting food on the table doesn't cost anything anymore, and they got whuffie from me just the same.
--
Power to the Peaceful
Although it would be nice if it took encoder inputs, which would make it useful for industrial applications.
To think that one's BSOD is now a MOBILE BSOD. Gives a new meaning to a system "crash!"
SRI Centipod is also based on VIA technology and already runs on Linux. It was demonstrated at Linuxworld 2003 in San Francisco. clicky
"Currently the bot runs Windows, but Roboteq expects that one of the first enhancements from the user community will be Linux support."
Considering that Linux support is added for just about everything, espescially when it's not wanted, why not skip this one just because they're expecting it?
so Roboteq wants linux users to purchase it for $495, and write their own code for it?
Isn't that the business plan for most companines deadling with Open Source software?
well that's what i was asking, i guess... Isn't there usually some type of incentive, or assistance given to those that are doing the work, i.e. the hardware costs? call me crazy, but i'm not going to 'donate' my time to a company so they can 'sell' a product including my work, and make even more money than they were before. when is the greater good of the community worth promoting corporate greed?
I was startled to see a new and unpredictable form of joke in these comments, implying that the robot would crash because it's powered by windows. You see, Windows crashes. So the robot will. Aha. So funny. I had not heard those implications before.
It was especially good when somebody implied that your screen would turn blue and the robot would begin a killing rampage. A robot rampaging and crushing Windows users is something I would have thought Linux users would cherish. Oh well.
Anyway, had you heard the joke about the cutting-edge Linux device driver for the new hardware?
No? Neither have I.
I happened to visit mini-itx.com earlier today. I was rather disturbed to read this article about taking a teddy bear, and disassembling and reassembling it as a 800MHz PC. It seems relevant to this topic, in a demented way...
This is interesting, but seems to be very little more than a companies advertisement for a product built for multi-applications and is a little off-beat for the usual non-commercial Slashdot article. Robot usage probably was not the original intent, but that is just my opinion. Although they have an impressive set of hardware specs, and could have used any X86 embedded processor, the controller seems to be someone's expensive golf cart controller set in a new application. Perhaps the most interesting thing is the 125 amps with no fan, but that is common on golf carts (and maybe Segway).
I've often thought that Erector Set construction kits would be ideal for custom robots. Unlike Legos they are strong. They are various strips and plates of metal with holes, most of them L-beams, that you bolt tegether as needed. They were sort of the Legos of the 1940's. My dad had a set from his younger years and I used them for science class projects that needed sturdy platforms.
But, the company bellied up. Maybe some Chinese firm will resurrect them.
Table-ized A.I.
If you don't want to spend the $500 and want a more affordable solution, try buying some surplus hardware. For example this page has stepper motors and controllers at reasonable prices.. htm
http://www.alltronics.com/stepper_motors
And instead of a VIA board you can get a cheap
80186 board for $69 from http://www.jkmicro.com/products/flashlite186.html
Okay you can't run Linux on that board but with 1 MB of memory and 33 MHz you can do lots with a robot.
Woa! You almost forgot the SCO license fee there Red Rider!
Now your bot just exceeded $1,000..
(Still cheaper than a broad if all you want is a beer getter)
VIA's Mini-ITX boards might have low power consumption while operating but they are total shit for power management. They leave practically everything they can powered up even during S3 suspend when only ram is supposed to be fully powered. Even when totally powered OFF they operate USB, SuperIO, and PS2 ports at FULL power, no matter what the OS asks for, totally failing their own specifications.
:) At least VIA has some truthful advertising on one point: the board is indeed a very small 170x170mm!
The boards are capable of pulling up to 2A off of ATX 5VSB in flagrant overconsumption. Even if the power supply is capable of providing that current on 5VSB, providing it continuously without the internal fans operating (which they don't in suspend) could be dangerous. The ATX specification provides for an absolute maximum current draw way below what the average Mini-ITX board draws from 5VSB when idle or off.
This problem was identified over a year ago after the ITX boards were initially released; however VIA has never issued a single fix for it, although it's apparently not the hardware preventing the fix from being implemented! Needless to say, their customer support in the Mini-ITX department is also shit. Is it so bad to want the board to work as advertised?
Luckily, I haven't even gotten into the processor or video driver issues, which many people see as an even larger problem than the power issues.. To each his own
~GoRK
When it starts screaming, "Exxxxxterminate... Exxxxxterminate," that's the time to flee in your Tardus.
What a load of crap. That's $500 just for the motor controller. All you need is a few serial handshaking chips, and power amps, and some simple circuitry to pulse the motors at a frequency proportional to the speed you need (pulsing rater than voltage varying gives smoother movement). In short $500 is a waste. At the most there is 50 (probably only cost $50 in US) of hardware in there. You're paying the $450 because of the times they've spent building it.
My laptop's power cord isn't making good contact in the jack, apparently, and it takes a few minutes of wiggling it around to get a connection. It's getting frustrating and is getting progressively worse. As I understand, the AC adapter is soldered to the motherboard, and if that's loose, it will also cause this exact problem. Has anyone else had these troubles, do you have any ideas on fixing this, or at least what exactly might be the problem? Thanks, I'd rather not pay to get this professionally serviced if at all possible. It's not under warranty.
Driving them is non-trivial, too. You have to spend as little time in the linear region as possible, to avoid overheating. Yet the load is inductive and is fighting that. You need overtemp detection and current limiting. You need protection against H-bridge short-through. At kilowatt power levels, all that stuff has to be there. Somebody had to make that all work under adverse conditions.
Incidentally, motor controllers are usually run at a constant chopping rate but with a variable pulse width.
Lately i've been rearing my ugly head at this page. The author has schematics for interfacing it to the LPT port, as well as software examples. I built my own interface using a rapid prototype board.
For those too scared to venture into building the interface themselves, I found this unit for less than $20 bucks from a company called ACS out in Florida. Unfortunately due to economic conditions, I cannot even afford that, but great looking product guys.
$500 is tooo much, i'd rather just interface steppers through the LPT and save an extra $380.
axiom.dll sucks
:)
A direct attack on my person, targetting my homepage? Whoa, I'm becoming famous
.: Max Romantschuk
Some people here have done something like this called beobots, a mobile robot platform with all design specs available through CVS. It's based on a highly adaptable and biologically based vision toolkit.
He who controls the past, commands the future... He who controls the future conquers the past.
Parallax sells (http://www.parallax.com/ ) pic based microcontrollers designed for hobbyists - they are great for robotics. They are fairly expensive for microcontrollers, but you could still do a hell of a lot more than this POS via will be ever able to for $500 and the power consumption will also be much less. Then again you could always just wire EDM yourself a small wankel rotary, hook that up to a generator and run off nitromethane...
http://www.motionmountain.rg3.net/
has been considerably extended. It adds a new section on nuclear physics that includes the story of radioactivity, a MRI scan of humans making love and the dream of grand unification. The present 14th revision also adds more figures on special relativity, a short explanation of k calculus, the paradox of the relativistic submarine, a photograph and description of how some caterpillars shoot away their faecal matter, a photograph of a basilisk running over water, Stowe's periodic table of the elements, a beautiful picture of the analemma, references to monadology, the question why birds are not usually seen on high voltage lines, the question whether the moon is larger or smaller than the nail of your thumb at the end of your extended arm, the proof that Peary did not reach the north pole, and much more. Many photographs and drawings have been added, the graphical presentation reworked and several hundred challenge solutions have been added. (Tell me which ones you want next!) For all fields of physics the newest research results and the main unanswered questions are presented. The text remains a structured walk through classical physics, relativity, quantum theory and unification. In total, the text aims to give an overview of what motion is and what it can effect. The accent on surprises and thought-provoking puzzles has been kept. It will please both readers who open the text at random and those who read it systematically. Thank you to everybody who has sent criticisms, corrections and suggestions. Enjoy!
Bruno Bezerra Lima
P.S. In physics texts, figures take the biggest part of the budget - both in time and in money. Any help on (preferably colour postscript) images to be added to the text (with permission of course) is much appreciated. Such help will allow me to concentrate on the writing and make sure that the text can continue to be free for all readers throughout the world.
Actually, in the ARL (Algorithmic Robotics Lab) at RPI, I'm working on a research robot based on the Via Epia ME6000. In fact, we've just switched away from the Ampro Encore PP1 (a powerpc board) because of Ampro's complete lack of support and general flakiness of the board. So far I'm reasonably pleased with the Via board, though we'll see how it works out in terms of power consumption (though even with conservative power usage estimates we still expect to get 3+ hours, down from about 4.5-5 with the powerpc). It was certainly trivial to get Linux 2.6.0-test7 up and running on the board, though I haven't tackled setting up the onboard firewire (for our camera) yet. From what I've seen so far, it's hard to beat the sheer functionality you get with the Epias, for the low price (we paid about $120 compared with $1000 for the PP1 with baseboard).
More info on our project here.
i think that robot will kill anybody it see if you dont install its driver currectly ! :) like VIA chips !
well, at least they aren't outright saying that they don't want linux to run on it.
it's 495$ for HARDWARE(the motor controller), which is perfectly understandable(the software being available for download, of course, you need a windows license around.. and it's not much use without the hardware).
world was created 5 seconds before this post as it is.
Unless the bot is going to do some image recognition or running advanced AI this is so overkill it isn't funny. Plus it will have a runtime that is insanely short for any real research to be done.
I made a "overpowered" robot 5 yearts ago with a baby 386 motherboard and the 1st generation wireless networking. and even with the AI I was messing with and basic image recognition on linux with a B&W quickcam, and I still had lots of processing power left over.
I was getting almost a full day runtime from the bot (depending on how much the motors were on. if you ran full tilt or worse started and stopped repeatedly you could kill the lead acid on board in 3 hours.)
and I spent ALOT less on my robot... including motors making my own motor controller and finding the 1st gen 802.11 wireless isa cards.
and they leave out important things like sensors...
Do not look at laser with remaining good eye.
Have a look at this Libre Hardware/Software design: PINO Bipedal walking Robot -- and Free as in Freedom from top to bottom.
Ummm, there has never been a time that you could not buy Meccano Erector Sets
Unless you count the time before 1990. At least, according to the article you linked to. That's the year Meccano France purchased the "Erector" trademark and started selling Meccano sets in the U.S.A. marked "Erector Meccano".
Though the original Meccano factory did shut down, other factories in France and Argentina continue to build and sell the sets under a different company name: "Erector".
Again, this is wrong. In the early 1900's the Gilbert Toy company (a competitor to Meccano) captured much of the U.S.A. market with it's "Erector" sets, according to the article. There was for a few years a Meccano factory in New Jersey (but it certainly wasn't "the original Meccano factory"), but it lost out to Gilbert Toy. For most of the century, Erector had the U.S.A. locked up, and Meccano had the rest of the world.
We've been running a contest that uses similar hardware. Take a look at http://maslab.lcs.mit.edu. Our platform uses a slow (but more power-efficient than the epia) Natsemi GX-1 (300MHz x86). We've built a custom robot controller board (4 motor drivers, support for quad-phase optical encoders, analog and digital i/o, etc). We're not mass-producing them, but should have a few extras at the end of January. Should be much cheaper than the $500 those folks talk about.
We use the CPU power to do vision processing and allow autonomous operation. The versatility of the ORC board allows teams to build really cool robots.
We used Linux last year and this year we're trying Windows XP Embedded with C#.
If you're in the Boston area, come check us out at the end of January!
I am in the process of building a prototype based on stepper motors and stepper controllers. The controllers cost a little, but eliminate the feedback requirement. These controllers can be daisychained on a single serial port. See: http://www.stepperboard.com/ The commands are simple 1000G to go 1000 steps clockwise and -1000G to go reverse. Speed is adjustable as well as a programmed speed ramp up and ramp down. I am using Redhat with a pair of webcams talking to the controllers thru the serial port. THIS ROBOT IS AUTONOMOUS... NO HUMAN REQUIRED.
Yeah... Since the interface on the controller they sell has RS-232, it's not like you'd need to write device drivers... just open the serial port and read& write the right values to it. They don't elaborate on the specific signals in their faq, but it does contain enough information to make me believe that such control would be fairly trivial
http://www.roboteq.com/faq.shtml#serialfaq
The Digital Sorceress
... check out the Open Automaton Project. This project includes electronic circuit schematics, software and documentation you can use to build an intelligent PC-based mobile for home or office environments.
The prototype is also based on on the same VIA mainboard as the RoboteQ robot.
it is VIA-based. And VIA will not let you
know what you are buying when you buy their
products. You see, if you gave customers
your datasheets, they might be able to actually
use the products, which could lead to all sorts
of legal complications. It's much better to
just sell them a product they can't use.
-I like my women like I like my tea: green-
While there is something to be said for scrounging parts and banging together whatever crap hardware I can find to do the job, its not a very scientific method. Being able to design something that is replicatable is a far more worthy project -- otherwise all your design work gets locked up in -your- project and crap on everyone else.
While this does not mean this hardware = solution to all robot problems, it is a pretty robust setup. The fact that it can handle everything from battle bots to personal transport is pretty impressive. Robot! To the mailbox with me!
This controller uses only two fets per leg of each H-bridge. The leads on the mosfets these controllers use are only rated to 70 amps each. Even so, the RDS-ON value for the fets really limits the current to much less than they quote. The 120 "Smart Amp" (whatever that means) is really only good for maybe 15 seconds. I would rate it closer to 80 amps continuously.
If you are looking for a high power, inexpensive motor controller - check out the osmc project! It uses four fets per leg. They also have the MC1, the goliath of inexpensive controllers. Rumor has it that a 100KW single direction controller will soon be released by these same folks for the price of their MC1. Much more powerful and easy to interface.
Mind you, it is pretty fancy. :)
For those on a budget and would be using Radio Controllers with PPM modulation, or feel like creating PPM modulated signals out the parallel port (or with a bit of PIC magic), Schulze electronic speed controllers would be a good candidate.
Granted, they're not as intelligent since they don't have programmable serial ports and such, but they do allow you a much greater range of battery and motor choice. Schulze also make sensor and sensorless brushless motor controllers, their latest capable of delivering 180 amps!