the crypt

--Well yes and no; you'd be amazed at how many web forums have gotten it wrong! Yahoo springs to mind.. ;-)

I've been lurking here for the last year or two, but while my robotics hobby has stalled, my intent to pick it up again remains. I've actually been welcoming the slowdown because I just don't have time to keep up with such a busy newsgroup.

I lurk in comp.graphics.algorithms, too (and I used to read 4 or 5 other graphics newsgroups, since I was working full time on that stuff and it really counted).

I'm curious but a bit behind the times, so would you mind telling me what these better forums are?

I just wanted say something to encourage those of you who are still active here to keep up the good work. It's certainly appreciated by me, even if I don't always say anything. It's somehow comforting to know that when I do start up my projects again there'll be someone around to help me with the many obstacles I'm sure to encounter.

- Owen -

" snipped-for-privacy@users.sourceforge.net" wrote in message news: snipped-for-privacy@z14g2000cwz.googlegroups.com...

I'm flattered you remembered.

My (unpoplular) premise in that post was DARPA GC had taken a great deal of the money and energy out of the community. Other's suggested it might be the spread of venues available, so I did a bit more data gathering. I added all the posts for c.r.m. with the Seattle, PARTS, Megabitty, MARKIII and OSMC lists, and came up with these totals:

While the trend is down, it is not clearly as spectacular as considering c.r.m. alone, and that would tend to support both hypothesis: the interest is declining, but also becoming more diluted by the increase in forums available to post.

I've been a software engineer ( or just plain programmer ) for over 20 years now. Built a few robots that were 90% kit based. Just started building my first one almost totally from scratch. So, I can really get into the software, challenge of working on vision recognition, etc. The MOST fun I have had recently working on the bot is ... drum roll ... cutting threads into the aluminum extrusions with a cheap tapping kit I got from Sears! I got the biggest kick out of doing that and then seeing the parts go together into a fairly professional looking chassis ! For me, at the moment, it is the building and creation of something that I just can not find elsewhere. Later will come the satisfaction of seeing it perform the way I intended - hopefully !

Just my 1.5 cents worth ...

JCD

No problem.

I'm just using the built-in color correction. To stabilize things, there is a setting to disable auto color adaptation and enter fixed settings. It also helps if you convert the image to HSL or HSV

On Linux, you might be able to use/modify the Philips WebCam (pwc) driver to obtain pixels in the native camera format; but this is of little interest to me.

Later, Daniel

Another explanation, at least for the US is the economy. Times were pretty darn good in the late 90's and peaking around 2000. Then it sort've went bust, at least for a lot of techies which make up this hobby. So instead of a little extra leisure time and a little extra cash to spend building robots, those same people instead are working the second job at night trying to feed and clothe their kids. Or working extra hard to try and not get laid off. Or were laid off and then went on the job hunt - no time for anything but economic survival.

-Brian

And yet here I am (next sunday) looking for something interesting to talk about. Go figure.

But like you, I didn't go to dpa's site, because I have gone there many times before. Figured I knew what he had up there. So I looked. I hadn't seen it before on his videos, but I had seen that behavior before in person.

Just a question...

Was the mucho dinero spent on the high end inertial components worth it compared to the lesser units out there?

Mike

Yes.

Well. maybe I should ask, what lesser units?

For the balancing robot, I spent about a year working on my own IMU, which I got to work reasonably well, but not really well. Not as well as a pot with a stick dragging on the floor At that time I decided that I was spending a lot of energy working on a problem (mastery of extended Kalman filtering) that was really tangential to the problem I was actually trying to solve. So I bought the single axis IMU and went back to work on the part that I was really interested in, the robot.

I have several friends that have gotten home-brew IMUs to work reasonably well, but not really well. For example., their robots don't appear to be able to handle the rough off-road terrain that nBot has been able to. The difference might be our control code, but I don't think so. I think it is the sensor.

For jBot, the cool thing about the IMU it uses is that it includes a

In honesty I haven't yet seen another robot that can do the same for comparison, so I don't really know the answer to that question. Although I've been trying really hard to get Randy to build (or better yet, have built) a similar platform...

cheers dpa

You know, I don't think it is as simple as that. (Forgive my biased opinion, but I truly believe:) The IsoPod(TM) was a $99 microcontroller better suited to robotics and motion control than anything industry had ever created. While there were notable exceptions, overall sales to hobbiest have been dismal enough I've discontinued advertizing. I think my HC11 boards did more for robotics, as I am fairly sure the MIT Handiboard was inspired out of my product line.

I think BattleBots had more to do with the 2000 spike in robotics, and

I'd like to think some new product could open up new worlds. I strive all the time to find them. For instance, it seems Bob has hit a homer with his Maxsonar-EZ1. The innovation is making it half size. But will it make a new wave of interest in robotics. I have my doubts.

You (Gordon) and I discussed a new sonar two or three years ago. We wondered if mounting them on a cheap scanning motor wouldn't be a good idea. I wondered if making them a tenth the size would open new markets and increase uses. So it isn't that the idea was missing. It just seemed Gerald Coe was doing a good job with his products. Clearly Parallax saw the niche as well, and did the PING. Gerald has responded with some new innovations, SRF05, SRF10, etc. Now Bob has announced. Still, I don't think this new wave of sonars will cause a new wave of robot construction.

Likewise, the CMUCam was an exciting new entry. But how many robots really make use of it? I'll bet there are10 or even 20 times the number of CMUCam's owned, than there are CMUCam's actually mounted on robots. I know mine is in a box. I even have worked out what I think is a viable plan to have predictive line following. I have written the interface to send and capture every camera command.

Seems to me the CMUCamII was a move in the opposite direction, bigger, more power hungry, harder to use (odd changes in communications vs. extentions and additions). So, why will a 25% reduction in the cost of a new "vision" system under the under utilized CMUCam's make for a new wave?

It's something else. It's a deeper driven wave than just waiting for the introduction of one product.

As I've already said, I am constantly thinking on how to make significant improvements in peripherals and sensors, and I'd like to believe something like those could energize a new wave in robotics, (See our 'Pod's, see our $29 ARM, look for our new high end H-bridge, look for our new low end H-bridge, look for our new Differential Drive Module, look for our new line of sensors... etc.) but I don't really take that opinion too seriously, because I've been proven wrong before, and often wrong lately. It's something else.

I wuilt a balancer with a "bare" analog devices accel, and and a silicon sensing systems gyro, which performed moderately crappy with about 20 minutes of coding.

At the last employer, we were using BEI systron gyros for sub degree sensitivity.

I can spend anywhere from 2K for an o-navi, 4-5K for a Rotomotion, 10K for a Crossbow, or 26K for a BEI.

I wonder how good is good enough to keep robots from tipping over on rough terrain. Have you done any studies on known terrain, comparing IMU data to reality, and calculating how poor your IMU can be?

Mike

You're right; it's not as simple as that. But newsgroup postings aren't great at providing a platform for complete analysis and reportage, and for something as complex as consumer buying patterns, no one thread will cover it all.

One part of the equation that's missing is how well the product is marketed to its audience. What the product had to offer, and how the market perceived its value, don't always coincide. Betamax was arguably the better tape format, but it lost out because its creators were convinced that it would be the dominant player simply because it was better. As we all know that's not how the story went.

What keeps people in the BASIC Stamp camp, or using 8-bit processors, is not the capability of the hardware, but the availability of examples and other support materials. Parallax has built an entire business not on a microcontroller, but on support for that microcontroller.

Absolutely it did contribute, but fads are like that. Robotics has largely followed the hobby electronics trends. As someone in this biz for a while, I'm sure you remember the ups and downs. As I've said before, this is a natural cycle, and I've been watching it for the past

I also have my doubts, though I wish Bob's product well. Gerry already had introduced smaller units, though at something of a premium price, and to me they didn't seem to be barn burners (but, I don't have insider knowledge here, so maybe they were). I'm not sure size has ever been a compelling issue with ultrasonic sensors, as the SRF04/08 were already pretty small. Or even the issue of accuracy. The logjam today is the lack of suitable examples to use what's already available.

I agree, because these products are not revolutionary in concept. They are incremental improvements of what's been out for 6+ years. It's becoming something of a commodity business, which is largely played by the price card. The winner is the person who can outlast the competition, given smaller and smaller margins. It's not a business segment I'd be interested in pursuing, I can tell you that.

Like the shortcomings with sonar, the problem with vision is the lack of example code to make it fully functional to the average experimenter. There really is a dearth of examples. I have many things competing for my time. I'm not likely to mess with some new piece of hardware unless it offers at least a degree of plug-and-playability.

I'd rather see lots and lots of examples, so that I'm not stuck reinventing the wheel to do common -- and not so common -- tasks. These sound like great products, but how fast can I develop something from them? Are there ready-to-go samples that will help me? Preferably not ina language I'm not likely to be familiar with. The best car in the world isn't worth much if isn't filled with gas; roads aren't as useful without a map. IMO, the actual product is maybe 25 percent of the effort in making something successful. The rest maybe something like this: 25% user education and support, 25% advertising, marketing, and other promotion, and 25% sheer luck. YMMV.

-- Gordon

Based on those numbers, and your limited sample sources to compile from, there may well be an increase in popularity in robotics, if you consider robotics related posts on all the forums available. Granted they are popular ones, but I know there are many others out there. It may be true that

I would be interested to see how the projections hold up as well. I notice a lot of traffic around holidays, as the young'uns have some time and money on their hands.

Another intersting source may be robots added to robots.net on a monthly basis over the years since its inception, though I suppose if it is being watered down with a profusion of robotics related material, it should see a decline as well.

We need to ask the computers that track all our electronic comms that the gov. runs what they show in terms of the volume of the word robot.

Mike

Hi,

I have more data now and have to modify my original observation and question.

One of the videos that I posted to this list has no other links, so any accesses to that video must be coming from this list. And over the last couple of weeks there has now been about 50 hits on that link. After the original posting there were two.

This is pretty interesting. The "normal" pattern of access that I have

seen are a sharp and almost immediate burst of accesses followed by a long exponential decay. Here's a plot of accesses for a posting on slashdot for the same website.

In that case number of hits rose to a maximum in just under 4 minutes from the time of the posting, to a peak of 78,000 hits per hour. Now that's not typical for a robotics forum, but the response curve is typical.

I haven't yet plotted the accesses from comp.robotics.misc, but it appears to me to be a very different sort of animal. It's more like the link "diffused" through the internet over the course of a couple of weeks, rather than the sharp spike and exponential decay seen previously.

Maybe what the data show is that information spreads from this net group in a fundamentally different pattern than some of the newer styles of internet group organization.

regards, dpa

I'm not sure slashdot is really a very good measure of genuine interest. Instead, it seems to be more of the equivalent of a riot that hops from web site to web site as new articles make the front page. Leading up to the DARPA GC, our Insight Racing web site was mentioned in a slashdot article and the hits on our site soared sky high. Usually we can expect a certain number of queries to us asking questions and so forth and the ratio for hits to queries is generally pretty constant. However, even though the slashdot article generated a huge number of hits, the queries did not rise in proportion. So I think slashdot traffic is composed primarily of "low quality" hits and do not necessarily represent genuine interest in the target web site.

Again, I think slashdot is a bad example.

-Brian

Hey Brian,.

I guess I did not express myself very well. As I said, the slashdot events are NOT typical of the sort of hits we get from postings on a robotics forum. But the SHAPE of the response curve is typical of all types of events we get from postings on robotics forums like robots.net, the dprglist, and the Seattlerobotics list.

I just didn't have a handy graph of the hits associated with postings on robots.net, the dprglist, or the seattlerobotics list, to use as an example. The slashdot graph I used was supposed to illustrate the shape of the response pattern, but that's all. I was not trying to compare a "slashdotting" with actual interest in robotics on various robot forums, and the sort of webserver events that they may generate.

However those robot forum events DO produce the same sort of sharp spike of access followed by a long exponential decay as illustrated by the slashdot graph. The slashdot data are only used to illustrate the shape of the curve, which seems remarkably constant for different types of news groups and forums. I have seen this pattern many times. Here, for example, is a simlar pattern that lasted almost two weeks, for a posting about the two-wheel balancing robot on the European technical website "buzz bazooka:"

It shows the same sharp rise and exponential decay over a matter of many days. This is the same pattern I see for links posted on most robot discussion groups, even for sites without RSS feeds.

However, the pattern generated by the comp.robotics.misc group is unusual, and I have never seen it before. That's the point that struck me. I think it looks more like diffusion.

best regards, dpa

Brian Dean wrote:

I think you came across clearly and I got the point of what you were saying, I think. I was only clarifying that hits alone do not necessarily represent interest, especially from sites like slashdot. I realize you were making a comparison w/regard to the pattern or timing for when the hits arrived, response vs stimulus. It may take a little while for the usenet post to propagate and actually be available to the news clients, and that varies depending on what news servers folks are using. Whereas a web based group generally gets updated synchronously - the posts are available to everyone in the group at the same time. Not quite so with usenet. Also, people's reading habits may vary for usenet than with mailing lists / web based forums. I know that I generally don't check this group daily. Sometimes I do, but there are times when a week or more goes buy if I'm really busy.

Regarding actual interest, it is hard to gauge interest simply from hits - I think slashdot proves that point. People will generally click on links as the read the posts - doesn't really mean they are terribly interested. Put it in front of them, they click on it - some folks make a living doing this with banner ads, etc. For many folks here, this stuff is just chewing gum for the mind. I just happen to think that raw quantities of posts to usenet or even other forums is not necessarily representative of widespread interest, or lack thereof, in this hobby. Probably a better measure would be sales of hobby robotics books or other enabling technologies. And with the rapid increase of titles at the bookstore over the past few years, as well as the many hobby robotics parts web sites that have been popping up, I'd have to conclude that the demand must be there, in spite of Randy's argument that the DGC has suppressed interest since 2002.

-Brian

Howdy Brian,

I'm sure you're right about the sales data. All I have access to is the web data from my own server, admittedly not representative, but which seems to show some significant patterns.

As an aside, the number of hits on the robot parts of the server have increased steadily since they were first posted in the late '90s. I routinely get more hits today than ever before. For example, the data on the slashdot graph was from 2003, when the site's nominal rate, before slashdotting, was about 400 visits a day. i.e., that was the background rate. Today that rate is about three times that, more like 1000 visits a day of "background."

It's hard to to know if that represents a real increase in interest in robotics, or just that three times more people are now using the internet than in 2003.

But I do think, based on the patterns that I've seen and herein described, that descriptions of C.R.M as a "wilted corner of the internet" and a "crypt" may not be accurate!

best regards, dpa

Brian Dean wrote:

Excellent! I've hung out here for years and am on a first name basis with most of the regulars. It's good to see that it's still alive and well and that it's demise has been greatly exaggerated :-)

-Brian