Mr. ( or Mrs. ) MLW has caused quite a stir. I am concerned over many of his statements, but one in particular lies nestled deep within a thread. In that thread, the widely published and highly regarded Gordon McComb posted " In any case, many of us have specifically mentioned using microcontrollers WITH a PC, so it's hard to imagine how you have missed these discussions. " to which MLW replied " I haven't missed them, I disregard them."
That sums it up right there. MLW is not asking for advice, he is simply grandstanding an opinion. This is USENET, he is allowed.
We as experts with no doubt several thousand years of collective advice have cautioned him that he is on a path of heartbreak and misery. He dismisses us.
Cool, I am down with that. Chalk up another sale for ITX. That in the very least may help bring the price down should I ever feel like embedding a PC into a robot, which I am disinclined to do for a personal project ( did it on too many professional projects, all with support microcontrollers, but that is a different soapbox.)
1) I would love to see MLW succeeded. We all need a $500.00 robot.
2) I would love to see what MLW defines as a robot. Maybe what we think a robot is, and what he thinks a robot is, are completely different notions. It may be that for HIS notion of a robot, HIS methodologies are completely acceptable.
Yes indeed. Like a crypt. Just wish your cohort Dennis would show his mug from time to time. There are many theories floating about as to why things are dead here. I personally think it is a matter of diffusion. When I had my SWAP site, (site without a purpose ), there was very little data on robots on the web. USENET was a good discussion forum and knowledge repository. These days, the knowledge is not centralized. It is subdivided and decentralized.
Yes, it has been quiet. I can't speak for others, but my limited participation in mlw's threads, were when I felt something needed to be said, not for his sake, but for the sake of other newbies which might have been mislead to choose a design method already known to be lacking.
More power to him, if he wants to use an audio channel (real of software faked) to a bunch of loose analog components to run his wheels. However, his asking for, then completely "disregarding", good advice, with a confrontational tone of arrogance, and the impeccable references of having worked for a robotics company in the mid-80's (that immediately failed... hum), could cause a newbie to go astray down a dark and frustrating path. I always feel that sort of thing needs to be challenged for the honest-lurkers sake.
But back to being quiet. I know I get feed up with a lot of the bad advice I hear around here, and leave for spells. You've been noticeably silent lately too. I notice Gordon has been gone for ... I guess it has been a couple months. I know I've missed him. I hate to see his return "disregarded" right off the bat.
I figure mlw will get into this, figure out sacrificing 25% of his PC for satisfactory control of one wheel is just too much. Now, I don't see him coming back to admit that. So after he mlw drives me, you, blueeyedpop, Gordon, etc., off (all published authors in the field) it will again get really really quiet. (Where's DLC anyway?) I kind of felt that a flood of bad ideas was what caused the last quiet spell anyway.
It's been dead here? I've been gone for a couple months, come back, and it looks like the place has been hoppin' all along. Oh, well.
I've always postulated what when robotics goes mainstream, the hackers look for the next neato thing. I accept the arrival of robots like Roomba with some trepidation. It's great to see someone actually succeed with a vacuum cleaning robot, but how many folks are building vacuum cleaner robots from the ground up now? It's definitely a time of transition.
I think the way to keep things alive is for us regulars to start some threads from time to time. (But hopefully not just threads that promote only opinion and argument. That's just my opinion of course.) Really, it's up to us to keep the flames going.
I'll knock on the old noggin today to come up with one or two subjects, and post later tonight.
I'm fairly sure Dennis is busy with small children and earning a living. Still, a chastising email sent his way is probably called for.
Me, my passions alternate between robots and beer -- what else is there, really? I'm swinging back to the robot end of things at the moment. I WILL get a teleoperation package for the masses out if it kills me.
I like to do a little brewing myself now and again. Robotics is a great way to fill the time while waiting for a good batch to finish fermenting. What are your favorite types of beer to brew? I often dream of building a robotic brewing machine, but then I am faced with a Budweiser and remember that the whole reason I brew is because robo-brew and robot vacuum cleaners are very similar: they both suck!
I wouldn't call it "grandstanding," but I would call it information posting. And when I did ask a very specific question, the Pro-microcontroller mafia took over the conversation.
One of the reasons why I dismiss the advice is because it is not a fact, it is an opinion. You are "outraged" that I disregard your opinion, but you fail to understand that you are, in fact, disregarding my opinion.
And while you guys get all pissy about it, I have tried to debate from a factual informitive stand point, and you guys have called me ignorant (or worse). I have over 20 years professional experience building complex computer systems. I started out doing hardware and moved on to software. I built my first computer in the 1970s, and that was for a robot. I have a very good about what I am doing an the requirements.
If you want to debate techniques, cool, we all learn. 1st be explaining it and 2nd by debating it.
Of course if you all know it all and agree on how things can be done what is there to talk about?
I am a true robot hobbies in the sense that I don't earn a living from electronics, programming and thus lack the expertise that you guys have. In that sense I am not much different to a newbie. And to be honest much of the advice is too vague to translate into any useful action.
To get a robot up and running without years of training in electronics and programming you need help and that means anyone who wants to help *or sell their electronic boards* to a newbie have to provide an easy way in.
Back before microcontrollers or Isopods when the only option was a collection of ICs people such as David L. Heiserman wrote his book "Build your own working Robot" with explanations anyone with a basic knowledge of digital circuits could understand. It was also self- contained. You didn't have to scan a hundred sources to get sufficient information to actually build it or use it as a template for your own robot.
Your isopods look very nice but as useless to me as a catalogue of digital ICs would have been back in 1976. I have no idea how to put them together or what they can or cannot do. Now I am sure the information is there scattered over the web or hidden in the brains of experts but I really don't have the time or resources to pull it all together.
But it would need to be controlled by a PC even if it required special hardware modules for real time operation because the stuff I am doing on the PC will not run on a PIC. And if run on Windows or Linux a simple shell provided so those with limited programming knowledge can do neat stuff with C or C++ without having to be OS gurus.
Dennis has a new daughter, a fairly new full-time job, and since writing the book on oopic, spends a lot of time as resident expert on the yahoo oopic forum. Also hangs out on the yahoo front-range robotics forum.
Why, John, I've always had a great deal of respect for you, and never took you for a newbie. Also, I guess I picked up last thread you were in Oz? Guess I had you in New England (or old England) in my mind. Sorry.
If I didn't make it clear, I'll repeat it now. Your suggestion, how to make a robot using a 'Pod, is an idea I've taken to heart, and added to my life's todo list. It was a very wise and valuable suggestion to me. And recognizing we've changed away from the previous course of this discussion, I'm renaming the thread to reflect your point.
Apparently you are not the only one that feels this way, I am very sorry to say.
Okay. There ought to be a book.
But... in the mean time, if you go to our
home page, click the Downloads button at the top of the page, select the 'Pod Download page item on the resulting page, then page down on that resulting page to "Appnotes -Example Programs", you'll find a number of complete programs for 'Pod based robots.
The first complete system is marked Mark III - Mini Sumo Robot code. This was the first code I wrote for Mike-3 which took first places in line following and Mini Sumo.
Further down is Lynx 5 Robotic Arm, which is a complete canned profiled motion system for the Lynxmotion ARM, ready to be commanded by keystroke, or remote computer. Or commanded by parameters as desired. This L5 code was the basis of expanded code for S.A.M.M. the biped walker done by Dan Gates, that was promised in months, and instead walked within weeks of its announcement.
Next line down is the source for the Lynxmotion H3 Hex walker. This was a very early application of the IsoPod(TM). 20 independent RC servos are controlled (18 joints and pan & tilt on the head).
A couple lines down from that, marked PID Controls is the source code from Shnoz, Dave Peterson's first placed line follower, with also capabilities for doing the DPRG Quick Trip and T-Time contests. This was again early PID code, which was a contest winner, with Odometry included. This source went on to be the basis for BeerBot and Dr. Huff's M5 tank.
Above all that is complete code from Mike Keesling article on doing Inverse Kinematics for a Lynxmotion L6 Arm, a featured cover article appearing in Nuts & Volts magazine, that can control two arms at once off a single 'Pod doing all the floating point IK necessary for them to achieve a commanded position.
Along with these complete source codes are dozens of links to short program segments to hand robotic interfaces, such as cheap LCD displays, Devantech Compasses, by PWM output and by I2C, and Devantech sonar rangers, both the SRF04 by pulse measurement, and the SRF08 by I2C. There are also segments for turning timers into 32-bit Quadrature decoders (up to 6 of them on one 'Pod).
Well, there's plenty more, all gathered up in one place.
Short of a book, these technical links should let you feel the 'Pods aren't useful. We've spent lots of hours using 'Pods successfully, and more documenting what we've done. This isn't good enough for us yet, but it is a start. Some beginners might find it useful.
Yes, control by a PC is certainly possible.
...But 'Pods (circa 2001) are not like PIC's (circa 1990). And our new 'ARMs (circa 2005) are even more advanced than that. Do you realize you can get a 512K Flash, 32K Ram, 60MHz, 16/32-bit ARM from us today for less than $60 that is 1.3" by 1.0", takes less than 100mA, and has its own onboard regulators, RS232, SPI, I2C, 10-bit A/D etc.? CANbus is an option too I think.
(We haven't announced these yet. But their ready to go, as soon as we have some software to demo them. They are supposed to be our April specials, along with the very similar $29 ARM board. Maybe the CAN driver chips are extra, I don't remember for sure if they're on the production board or left open if you want them installed.)
This is (like) the one our customer has doing TCPIP scripted web servers over Ethernet (chip and magnetics wired wrapped on) already.
Oh well. We've still got a ways to go here. We tend to leave the PC's and OS's to others.
So, back to your idea of a book. I think we could do one with the following robots.
Mike-3 (Mini Sumo and Line Follower, contest winner) L5 (5 D.O.F. Arm) Dual L6 (Dual 6 D.O.F. Arms w/I.K.)
There's no fantasy in these. All these 'bots exist, and are (or were) operational.
Oh, almost forget Pete Gray's TankBot featured in DDJ. Pete would that be okay to add to the list? Anybody else have a working 'Pod robot they'd want to have featured in a book to suggest? I know there are a bunch of them in DARPA GC, and doing RoboMagellan mini challenges, autonomous aircraft, etc.
And I'm sure before we could finish a few chapters in the book, several more of ours would be operational, two which are imminent.
Still think you need a bit more of a comprehensive tutorial for absolute beginners. The docs assume a few things , which take a while to work out.
The Lpc2138 boards are ready then. Nice Just in time for the lpc21xx circuit cellar contest starting on the 25th .
Mikes robopede looked awesome , pity it would be rather expensive to build something similar.
The code for that was in small c?
That would be good.
The lynxmotion scouts look good bit like one of the walkers from star wars.
Maybe you guys should get that one out before the movie is released (may the 15th) and get a few videos out and circulating other than the exisitng one. Suggest to put a large isopod or servopod sticker on the front of the bot. Something like powered by servopod or isomax inside.
In particular... For some reason, the state machines which come very naturally to hardware engineers, seem to start out as a foriegn idea to software-only engineers. Several software engineers have said once they "get it" they say they can never look at programming the same again. All the sudden, all their tricks with flags, and variables, and conditionals, which they've essentially used to make hidden statemachines, stand out clearly as an inefficient way to do things.
It might be just something that takes time to catch on.
Yes. Production protoypes proven. Rest are kitted up now, ready for full production.
True, there were 12 motors, each with a big H-bridge, and two each with a processor. It did some surprisingly agile things. The sum was greater than any of the parts.
Yes, Peter Gray's TankBot was in Small C.
Yes it does a bit.
Maybe not too. I've got several university Physics Department Colloquim to give in April as invited speaker. And I've been sick almost two weeks now. Things are backing up.
Bill James put a vinyl set of lettering with something like that on BeerBot's new skin. I forget the exact wording. Something like "Brains by IsoPod(TM)" or such.
Hi Randy, this sounds very interesting, except for one caveat. The power and program space sound great for a more powerful robot, however at the same time, to build robot apps more powerful than the usual control and sense features, what is really needed I think is a "lot" more available RAM, ie data space for a lot of sensor processing, mapping, etc. 32K doesn't quite cut it. So, possibly a mezzanine RAM board might be nice. A multi-tasking OS would be good too, considering such a cpu could handle it.
For a couple of years, I have been scanning for a "next-step" processor for a robot. One possibility is going all the way to an ITX style board, but they're terrible power hogs, and unsuitable for any robot under maybe 10-20 lbs. The only other thing I've found appealing is the Rabbit, and mainly because the boards come with up to 512K of RAM.
No, at some point, even I'd go to an ITX. (You know it hurts me to even say this right?)
But we do have something ARM based with 4Meg RAM in prototype.
Yes, the Rabbit has some advantages... in a very Z8ish sort of legacy prone way. Certainly up til now, if I had to do an embedded Ethernet, I would have looked more closely at the Rabbit. But with the big ARM's just around the corner, with USB slave/host and Ethernet capabilities... maybe not just now.
Have you looked at the new 200MHz+ ARM processors like the MX-1 from Freescale w/BlueTooth accelerator among a whole host of things, or the AT91RM9200 from Atmel? or the LH7A400's from Sharp?