Here's the transcript from our second unreal-time chat conducted with BEAM robotics guru Dave Hrynkiw on Shop Talk, our conferencing area. We'll be doing more "Geek-in-Residence" chats in the near future with computer and digital culture luminaries. If you haven't checked out Shop Talk, please do. We give away cool prizes periodically to lure people inside. And, if you have any suggestions for Geek-in-Residence guests, please email me. - Gareth

The second guest in the Street Tech Geek-In-Residence Program is Dave Hrynkiw (as he likes to say, the last name is pronounced "Smith") of Solarbotics, a BEAM robotics kit and parts company.

BEAM is an acronym for "Biology, Electronics, Aesthetics, Mechanics" and describes a controversial field of robots were small, simple, biologically-inspired robots are made out of electronic parts, junked Walkmans, pagers, etc., and solar panels. The goal is to build autonomous bots using basic analog circuits for control instead of microprocessors and sophisticated software programming.

One unique aspect of the BEAM approach is advance the field through robotic games where humans are the designers and the robots are the competitors. Or as the BEAM maxim goes: "A human being is a robot's way of making another robot." Through this process, a slow robot evolution occurs where the bots grow smarter and adapt to the new conditions they're faced with.

It's a nifty idea. I've been lurking on the BEAM mailing list for the last few weeks. There have been fascinating discussions on building robot ecosystems, what you would need to create flocking behavior in a group of BEAM bots, and the idea of a BEAM automobile that would use survival instincts to avoid a crash.

For those who may be new to the BEAM concept, here are a few popular articles on the subject and two BEAM-related websites to give you some background:

If Tiny Robots Ruled The Earth This is a piece I did on BEAM for the ezine STIM.

Robots From Rubbish An installment of our pal Mark Frauenfelder's FRINGE column at Wired News.

There's A Robot in Your Future A piece covering several "fringe" areas of robotics written by well-known culture critic Mark Dery for Discovery Online.

Solarbotics Dave's excellent site with lots of bot pictures, info, and descriptions of Solarbotic's kits.

BEAM Tek A very well-done site that houses the BEAM FAQ, info on BEAM circuits (called "Nervous Nets") and solar engines, a glossary, and a link list.

So, after you've boned up on BEAM, please feel free to ask Dave some questions.

And Dave, welcome to Street Tech.

(If you get here before I do on Monday morning, feel free to introduce yourself and tell us more about BEAM.)

So Dave, give us a little background on you.

How did you get involved in BEAM?

What's your professional background?

How's biz at Solarbotics?

Where do you sell your kits?

You just did a BEAM workshop (at Los Alamos, wasn't it?). What exactly did you do and how did it go?

Hi Dave,
So how immune to cats is that cute little PhotoPopper? :)

Q1) Is there any reason that a 74HC244 would act any different than a 74HC14, they have hysteresis switch points ??

Q2) If I change the 74HC244 to a 74HC240, would I see all lights on but one??

tks dlw

The image in the first message looks very nice! Is this a standard Photovore using a plain old FLED/Zener type pair of solar engines?

I've never seen a pager motor really compared to anything till that photo! The motor is TINY compared to those transistors (which make neat references!)

Anyway, I was wondering what the best gear ration is for a 2 legged walker? I heard 1:18 somewhere, but that seems awfully weak to me, but I haven't tried it. I'm certain the gears won't strip easily with that ratio. I had two wonderful motors with gear boxes from a Minolta video camera lens assembly, but I stripped one of the gear boxes.

Also, what is the best way to build gear boxes from spare gears. Just drill holes in PCB and solder it together???

Richard Piotter

richfile@prairie.lakes.com

http://www.geocities.com/SiliconValley/Lakes/Richfiles.html

Hi David,

I built a nervous network walker from Miller's site. Now I will try and put together a solar engine to try and power the walker. Can you recommend a good design that can be made from the recycle bin?

That image in my initial post is actually on Dave's Solarbotics site. The URL is http://www.solarbotics.com/beam_2.html. It's Mark Tilden's BEAMant. There's a link to a much larger GIF. Part of the description reads:

Construction is based around a thin PCB chassis where a pair of recycled pager motors have been used for motivation. The neck of the robot carries a string of 10 photodiodes that produce between 3 and 5 volts to the system.

I'm sure Dave can tell us more about it when he gets here. Oh Dave...

Has anyone made one BIG solar engine with a powerful Solar panel that fires both motors, but uses photoresistors to adjust the power going to each motor so it travels in straight lines or smooth curves, because I assume the 2 engine Photovores fire out of sync quite often, making it spin in circles sometimes. I've actually never seen one operate in real time, but I assume this is normal operation.

Richard Piotter

On 4/27/98 4:18:22 PM, Gareth Branwyn wrote:
>How did you get involved in BEAM?

My sister sent me an article from the August '89 (? I think?) issue of "Scientific American" about this engineer that built solar-powered robots out of junk in his basement (Mark Tilden, of course). Struck a chord in me. I'm the youngest of 4 in my family, and had always relished in destroying my sibling's possessions, and rebuilding them in a new form.

>What's your professional background?

I'm a Certified Mechanical Engineering Technologist with experience using AutoCAD to design oilfield down-hole tools. Very related, wouldn't you say?

>How's biz at Solarbotics?

Keeping me busy! Especially with the recent exposure Mark Tilden has had on the Discovery Channel's "Robots Rising," and the Canadian issue of "Reader's Digest" ("Robots of the Mad Scientist" - should be in the US pretty quick).

>Where do you sell your kits?

So far, just thru our website and some of our products go thru Mondo-tronics. We're hoping to expand our retail/wholesale connections in the next few months.

>You just did a BEAM workshop (at Los Alamos, wasn't it?). What exactly did you >do and how did it go?

The Los Alamos workshops are an annual event (just finished ...#5...I think) held by the Los Alamos National Labs where Mark Tilden is employed. The event is an outreach program that is designed to get more of the local population & student body exposed to what it is they do at the lab, and hopefully encourage some of them to become scientists. It's a splendid 2-1/2 day event, where we helped over 80 kids & students (from jr high to college level) put together our kits and learn about BEAM tech.

My job there was to plant my flabby butt down on an excruciatingly painful fold-out chair and troubleshoot the robots that weren't working properly (Me gripe? Never....). My wife, Cheryl, manages the event (and very well too, if I may say so). We also had help from Mark Tilden, Richard Weait, Jan Frigo and Scott Martin and Rick Diaz - names that have all been seen on the BEAM mailing list.

With that said, onto the guest mail!

Regards,
Dave

On 4/27/98 7:15:04 PM, Richard Piotter wrote:
>The image in the first message looks very nice! is this a standard
>Photovore using a plain old FLED/Zener type pair of solar engines?

That's the great-granddaddy of the Photopopper kit. It's a "HBS" (Happy Birthday singer module) triggered, dual-solarengined "BEAMant," powered by a string of 0.5V, 0.5mA photodiodes configured much like solarcells generating 5V. An antique, but very classy... A Tilden design, of course!

>I've never seen a pager motor really compared to anything till that photo!
>The motor is TINY compared to those transistors (which make neat
>references!)

Ah, just wait 'till you see what we've got in store in the future. Possibly 4mm diameter pager motors - about 1/2 the diameter of the ones in the picture. Too soon to say when we'll have them, but I'm jealously guarding my samples!

>Anyway, I was wondering what the best gear ration is for a 2 legged
>walker? I heard 1:18 somewhere, but that seems awfully weak to me, but I
>haven't tried it.

Yikes! Yup, that is awfully weak. It boils down to the size of the motor powering the gears. On average, I find that about 1:70~ 1:200 is a decent place to start. Trouble with building a walker is finding CHEAP suitable gearmotors. If nothing else, you can always strip the brains out of a servo, and use it as a gearmotor - it's a proven technique.

>I'm certain the gears won't strip easily with that ratio. I had two
>wonderful motors with gear boxes from a Minolta video camera lens assembly,
>but I stripped one of the gear boxes.

You _will_ want to stay clear of worm-drive gearmotors. They won't "back-spin", and they are very easy to grind into mush. Good gearmotors (for walkers) should allow you to put torque on the output shaft and spin the motor itself. If you can't do that - look elsewhere.

>Also, what is the best way to build gear boxes from spare gears. Just drill
>holes in PCB and solder it together???

Hmmm. Haven't done it that way, but I like the idea! When I've built my own, I start by mounting one on a wire shaft, and build a wire ladder over it with the next gear. In general, this is a painstaking process, and isn't very strong, but your idea should be much stronger.

-Dave

On 4/27/98 5:43:38 PM, David Winchell wrote:
>Q1) Is there any reason that a 74HC244 would act any different than a
>74HC14, they have hysteresis switch points ??

Hold on - must find data book....

Not being a microcore expert, I'll give this my best shot (I like solar-powered bugs more):

Lessee - the 244 is a straightforward buffer. It'll take a signal and "push" it through with more muscle (but not amplifying it - that's another kettle of fish). It doesn't invert the signal like a 7414 would. If you compare the 7414 to a 74240, THEN, you get similar results, as the 240 will invert the signals just like the 7414 could. You could get the 240 to act much the same as the 7414 would in a microcore, but it won't be as easily influenced by sensor inputs and the like. This is because after each signal is pushed through a gate, it's been buffered again, and has lots of power behind, much more so than the 7414 would. Also, the 240 has 8 gates, where the 7414 has 6.

As for hysterisis, the 244 doesn't seem to offer much in that behaviour.

>Q2) If I change the 74HC244 to a 74HC240, would I see all lights on but
>one ??

Ah - should have read this first. See question 1 ;>

Actually, I would _expect_ to act much the same as a regular microcore, just not as easily influenced by sensor inputs.

-Dave

On 4/27/98 8:06:03 PM, Joe Knight wrote:
>Hi David,
>I built a nervous network walker from Millers site. Now I will try and put
>together a solar engine to try and power the walker. Can you recommend
>a good design that can be made from the recycle bin?

Mechanical design or electronics design? Mechanically, you're in for a challenge! My only solar-powered walker (SunScout - detailed on our website), is the most expensive robot in my Robot Jurassic Park. The two tiny gearmotors used for driving the legs are worth over $150 USD/each! You _NEED_ very efficient and high-quality gearmotors to build a solar-powered walker. Forget about modified servos - too much power required. Look in surplus magazines/websites for gearmotors that will run in the 20 to 100 mA range, with names like "MicroMo", or "Portescap".

As for the electronics, I once published the SunScout schematic to the BEAM reading list. I believe it's solar-power controller circuit is detailed in the BEAMTEK FAQ. Use "H" or "HC" series chips for that design (74H14 & 74H245) - they're designed for lower voltage operation (especially the "H" - 2V to 6.5V I believe).

-Dave

On 4/28/98 12:21:46 AM, Richard Piotter wrote:
>Has anyone made one BIG solar engine with a powerful Solar panel that fires
>both motor, but uses photoresistors to adjust the power going to each
>motor so it travels in straight lines or smooth curves

Not that I'm aware of. The problem with the approach you're describing is that each path using a CdS cell (photoresistor) has a minimum resistance of at least several hundred ohms. That's LOTS of resistance to a circuit that's trying to pump power thru a motor with only about 8 ohms resistance. Point is, even if it could direct the power, the minimum power of a CdS cell simply won't let enough power thru to the motor to make it turn. Now, that isn't saying there's ways around this - there most probably are. It's a sound idea, but somebody just has to sit down and adapt the theory to a working circuit.

>I assume the 2 engine Photovores fire out of sync quite often, making it
>spin in circles sometimes.

That depends on the amount of energy stored in the capacitor of the photovore. If it's a 0.33 FARAD capacitor - you bet, that sucker will do donuts better than a teenager in a honda civic in a snowy parking lot. But if it's only using a 4700 microfarad capacitor, it will only make small jumps. Our Photopoppers will cover 1/2" per discharge at best.

Regards,
Dave

Dave, for folks here who are unfamiliar with BEAM, can you bring them up to speed a bit?

What are some the main concepts behind BEAM robotics?

How does BEAM differ from other approaches?

What would constitute BEAM state-of-the-art at this point?

Are there practical applications for BEAM robots?

Are any of these applications currently being investigated?

Dave;

What brand and model of servos do you prefer to use in your walkers?

>>I've never seen a pager motor really compared to anything till that
>>photo! The motor is TINY compared to those transistors (which make
>>neat references!)
>
>Ah, just wait 'till you see what we've got in store in the future. Possibly
>4mm diameter pager motors - about 1/2 the diameter of the
>ones in the picture. Too soon to say when we'll have them, but I'm
>jealously guarding my samples!

Any recommendation to where you got those samples to keep me satisfied till (I hope) you start officially selling them! (=

also, I have these REALLY geared down boxes from old 70s clock radios. GOSH are they geared down. They are about an inch in diameter, and less than 1/2 inch thick. Has anyone found a use for these??? It almost did 1/4 turn with 200 turns of the shaft! It's probably between 800:1 and 1000:1!!! Is there any use for that???

>>Anyway, I was wondering what the best gear ration is for a 2 legged
>>walker? I heard 1:18 somewhere, but that seems awfully weak to me,
>>but I haven't tried it.
> >Yikes! Yup, that is awfully weak. It boils down to the size of the motor
>powering the gears. On average, I find that about 1:70~ 1:200 is a decent
>place to start. Trouble with building a walker is finding CHEAP suitable
>gearmotors. If nothing else, you can always strip the brains out of a
>servo, and use it as a gearmotor - it's a proven technique.

Any good way for those who don't have cash and are an hour from the nearest place that sells servos??? (salvage! :)

Richard Piotter

On 4/27/98 4:36:36 PM, Tim Tate wrote:
>Hi Dave,
>So how immune to cats is that cute little PhotoPopper? :)

I wouldn't directly know, as my wife won't let me get a cat (sigh). But I do have a big lovable dog that loves to play with them on the floor until she gets board. Then she lays on them.

(In short - they're quite robust!)

-Dave

On 4/28/98 10:25:39 AM, Charles Monica wrote:
>What brand and model of servos do you prefer to use in your
>walkers?

I've experimented successfully with the Futaba S148 and the slightly-cheaper S3003, as well as the Tower Hobbies TS53. All are modifiable in much the same sort of way.

On 4/28/98 1:36:04 AM, Gareth Branwyn wrote:
>Dave, for folks here who are unfamiliar with BEAM, can you
>bring them up to speed a bit?

Ah - good point. BEAM is a relatively new field of robotics where microprocessors are *optional*, not necessary. This is due to the simple, elegant electronics inherent in most BEAM designs. And since we focus on using solar-power and "recycled technology" (junk), our creations are relatively within the range of the average person to construct.

>What are some the main concepts behind BEAM robotics?

Build a robot that does something (doesn't matter what). See if you can "domesticate" it in the next version, to refine a quality it exhibited that you like. The neat thing about BEAMbots are that they're inherently CHEAP, and can be built for the less than the cost of a case of beer, so you can refine ideas quickly and relatively cheaply.

Also, by using solar power, we get away from batteries, which is the #1 cause of dead robots in the world. I have a table full slow (compared to batterybots) moving robots, that don't have an off switch, and that have literally been in constant operation for years. Longevity pays in the real world.

>How does BEAM differ from other approaches?

We focus on a "bottom-up" and "outside-in" approach, where we're using clever mechanical design to replace hundreds of lines of computer code, and simple electronics to govern the behaviours we want. It reinforces the "GIGO" concept of robotics - "Garbage In, Garbage out". You can't get a functional, robust robot using sloppy construction and expect software fixes to catch the bugs in you design.

>What would constitute BEAM state-of-the-art at this point?

Mark Tilden's "ROSWELL" walking robot. Although battery powered (and for good reason - people simply aren't patient enough to watch solarpowered walking robots), it is a 4 foot high, 4 limbed robot that uses a simple controller to govern it's walking ability. Mark's still working on the sensor package, and when it's done, it'll put many other past major robot "superstars" to shame. Truly a slick machine.

Mark's also working on a new series of solar-powered BEAMant Photovores that are so active and animated, you would think they're battery operated. These are truly slick little machines, and since other robots this size (3" square, 3/4" high) usually have dead batteries in 20 minutes, we're looking into marketing these as experimental platforms for hive behaviour for research institutes. Simply add light, and watch'em go...for days at a time. Try analyzing hive behaviour when you have to keep replacing batteries.

>Are there practical applications for BEAM robots?

Depends if you consider driving my dog insane a practical application....

In truth, there aren't any practical applications for *any* autonomous robots right now, regardless the technology. Now don't mistake a robot for a glorified remote-controlled car, like the ANDROS explosive ordinance investigation robot is. That's a complex machine, but not truly a robot. The Mars Pathfinder is about the closest example to a practical robot device, but even that had to have a great whack of instructions loaded into it's memory at the beginning of each day. Hardly autonomous, but was a great example of autonomous/tele-operation.

Another example of a practical example of BEAMtek in action is with the new "Pixelsats", or solar-powered BEAM satellites being researched by Mr. Moses & Ms. Frigo at the Los Alamos National Labs. Having self-orienting solarcells on satellites would solve all the problems the MIR station had a while back, no?

>Are any of these applications currently being investigated?

Well, there has been work done in the field of mine-explosion technology with BEAMtek. The idea is to dump cheap robots into a field and let'em blow themselves up on the mines they find. As long as they're cheaper than the mine, it'll be an attractive solution. But that's work being done by the Gov't. Who knows if it will bear fruit.

On 4/28/98 10:46:59 AM, Richard Piotter wrote:
>Any recommendation to where you got those samples to keep me
>satisfied till (I hope) you start officially selling them!
>(=

Sorry - company secret! Said too much already (consider that a streettech exclusive).

>They are about an inch in diameter, and less than 1/2 inch thick. Has
>anyone found a use for these???

Coffee-table coasters comes to mind...

>It almost did 1/4 turn with 200 turns of the shaft! It's probably between
>800:1 and 1000:1!!! Is there any use for that???

Yikes. Well, sure - why not? One idea I've played with is a BEAM Bungee-jumper. Have it hoist itself up a line, then plummet down and BOINGBoingboingboingboingboing... Need torque to do that.

Or use that torque to build up power in a spring. Go launch a BEAM patriot missile!

-Dave

Dave, BEAM is rather controversial in that I've heard other (let's call 'em CPU-head) roboticists accuse Tilden, et al. of ascribing the language of biological behavior onto "dumb" analog circuitry and coming up with fancy terms like "microcore" and "nervous net" to describe such circuits.

And by the same token, BEAMers seem to almost anathematize the microprocessor and are down-right apologetic when they mention them on the BEAM list.

Can you speak to us about this for a moment?

my truck won't do donuts...but here's what gets me..how does the more common 2-motor walker keep in sync (8 legger yeah go ahead and scoff but I'm hypnotized by the moving)?

Landon Mills

I took the rivets apart and re geared it to about 50-60:1, not sure. hard to tell. Anyway, i figure i can have a pulley on it and drive the pulley with the motor. that'll come around 70, right, nice for a small walker, right!

Anyway, I have a TON of old video camera parts, and I'm thinking the cassette guide would make a nice ram for a sumo bot! I could connect a high torque motor to retract it, and then the motor automatically reverses. this allows slack. The motor stops just before the snap lever, and when a photoresistor gets dark infront, the motor finishes it's return spin and snaps the ram! then it auto reverses and comes back, stopping before the lever again! Sound feasible. Probably require battery power though. Are sumo bots allowed to use batteries in competitions, or do they require solar panels, or does it matter?

Richard Piotter

On 4/28/98 4:50:51 PM, Gareth Branwyn wrote:
>Dave, BEAM is rather controversial in that I've heard other (let's call
>'em CPU-head) roboticists accuse Tilden, et al. of ascribing the language
>of biological behavior onto "dumb" analog circuitry and coming up with
>fancy terms like "microcore" and "nervous net" to describe such circuits.

Well, it _is_ a fancy term because there really isn't any other term suitable to describe what's going on with these analog circuits. On the surface it looks like a simple little chain of inverters, but the subtlies of it are far more complex. They don't have theoretical physicists at LANL looking at these thing just because of the pretty blinky lights!

>And by the same token, BEAMersseem to almost anathematize the
>microprocessor and are down-right apologetic when they mention them
>on the BEAM list.

Ok, let's look up "anathematize"...hold on...anteater (oops - too far)...Ah Ok. Essentially to "call bad". I think that's a mis-interpretation taken to the extreme. There _is_ a place of microprocessors in robots, but we simply haven't found a need for them in BEAM _YET_. Most BEAMer's (myself included) find we're having so much fun with all the different ideas we can implement with BEAMtek, we don't get up to the really complex tasks that would benefit from microprocessors. A good example of this would be a "horse & rider" walking robot. Use the microprocessor to prioritize tasks, communicate with GPS & relay data (the "rider"), and simply tell BEAMtek where to go and let it figure out the walking gaits and relevant obstacle avoidance (as the "horse"). I personally don't have a need to build such a device, but it's good to know that its possible combining the two technologies.

So far, we're getting behaviours and desirable results out of analog circuits that _usually_ require microprocessors. I think that the introduction of the PIC chips and other super-handy cheap programmable microprocessors have really made it too easy for people to slap one on a design and ignore routes the traditional IC logic route. It's much easier to throw in a few lines of code (hey, no wiring & extra soldering) in a microprocessor than to add an additional circuit, but in applications requiring robustness, BEAMtek is more than holding its own.

>Can you speak to us about this for a moment?

No. I refuse to. ;>

On 4/28/98 11:46:18 PM, Richard Piotter wrote:
>that'll come around 70, right, nice for a small walker, right!

One good rule of thumb with judging motor strengths for a walker is that the finished robot should be able to pull itself up when dangled by one leg.

>Are sumo bots allowed to use batteries in competitions, or do they
>require solar panels, or does it matter?

Sumo is definitely a battery-powered sport. BEAM sumo is slowly giving way to more mainstream rules, as BEAM sized sumobots are much larger & heavier than their international counterparts. Japanese (10cm square) or WCRG (9" square) are more popular, and I'm hoping to push BEAM sumo to fit into one of these more popular categories. Sumo is an easy way for people to come to a BEAM event and have something to "traditional" to compete with.

On 4/28/98 10:49:21 PM, Landon Mills wrote:
>my truck wont do donuts...but here's what gets me..how does
>the more common 2 motor walker keep in sync (8 legger yeah
>go ahead and scoff but I'm hypnotized by the moving)

I'm not sure what you're asking. Do you mean how the 2 motor walkers synchronize 4 legs? If so, the front pair of legs are connected to one crossbeam mounted on the front gearmotor which is angled slightly up from horizontal, and the rear is the same, but kept totally horizontal. The action is this:

Front Clockwise
Rear CounterClockwise
Front CounterClockwise
Rear Clockwise

It's really hard to describe in text. Andrew Miller's website (AM Innovations) has a very good graphical explanation of it.

why do you wait till all odd hours to get on?? as for Microprocessors (I know it's out of sync) I think we should use beam and Microprocessors in the pathfinder style robot

we know that if we were to put all the functions of the pathfinder into resistors and stuff..we'd need a truck on the moon. Like you said about horse and rider. let The processor give a general direction...let our beam stuff drive it there...and the processor tells it when to stop so it can do what ever. and wouldn't a walker be better on the rough terrain of the moon if you gave it a high step??

Landon

The AM Innovations URL is http://www.golden.net/~amiller/

BTW: Folks reading this may not be aware that there's a BEAM Webring at: http://www.geocities.com/ResearchTriangle/1142/webring2.html

It doesn't have all the BEAM sites on it (Mark Tilden's Official BEAM site, for instance). but there are some interesting sites on the ring.

On 4/29/98 1:11:33 AM, Dave Hrynkiw wrote:

>extreme. There _is_ a place of microprocessors in robots, but we simply
>haven't found a need for them in BEAM _YET_. Most BEAMer's (myself
>included) find we're having so much fun with all the different ideas we
>can implement with BEAMtek, we don't get up to the really complex tasks
>that would benefit from microprocessors. A good example of this would
>be a "horse & rider" walking robot. Use the microprocessor to
>prioritize tasks, communicate with GPS & relay data (the "rider"), and
>simply tell BEAMtek where to go and let it figure out the
>walking gaits and relevant obstacle avoidance (as the "horse").
>I personally don't have a need to build such a device, but it's good
>to know that its possible combining the two technologies.

The 'horse & rider' seems to be a really nice concept for practical applications, because BEAM works well for self-sufficiency type functions, but may not be as practical to use for higher order functions. I think a modification of the concept might be even better - Gareth gave the example of a fleet of imaging nanosats. Working independently, some problems exist with pure BEAMsats, like coordinating imaging, and transmitting data to the ground. So let's imagine a "sheep & shepherd." (Best I can come with right now.) Make a "shepherd" sat, that has a transceiver, command, & data buffers. Make "sheep" with a small area, hi-res camera, and a low-power, slow scan, short range transmitter. The shepherd gives the sheep commands to image an area, the sheep relay the image data to the shepherd, and then the shepherd relays it to the ground for image assembly. You still have some issues, like constellation spreading, but you are fairly immune to bit errors, and "sheepsat" failures. Sheepsats might even be more practical for missions like FORTE (a recent sat out of Los Alamos that detects nuclear tests), where you want to cover the globe cheaply.

>So far, we're getting behaviours and desirable results out of analog
>circuits that _usually_ require microprocessors. I think that the
>introduction of the PIC chips and other super-handy cheap programmable
>microprocessors have really made it too easy for people to slap one on
>a design and ignore routes the traditional IC logic route. It's much
>easier to throw in a few lines of code (hey, no wiring & extra
>soldering) in a microprocessor than to add an additional circuit, but
>in applications requiring robustness, BEAMtek is more than holding
>its own.

You just can't beat the REconfigurability of a microprocessor though. I guess every tech has its strong points. BEAM's counterpoint would be to just build another, since it is (or should be) so cheap.

And I really want one of those butterflys to put on my windowsill in my new office. How hard are those kits to assemble?

Dave;

What RC values have you found to work best in MicroCore walkers using servos?

On 4/29/98 2:08:12 PM, Charles Monica wrote:
>What RC values have you found to work best in MicroCore walkers using servos?

Lessee...it sort of depends on what you're using for batteries and motor drivers, but I've successfully used 0.1uF caps with 2.2 meg resistors.

-Dave

On 4/29/98 1:10:57 PM, Larry Smilg wrote:

>So let's imagine a "sheep & shepherd." (Best I can come with right
>now.) Make a "shepherd" sat, that has a transceiver, command, & data
>buffers. Make "sheep" with a small area, hi-res camera, and a low-power, >slow scan, short range transmitter.

Perfectly viable and entirely correct. I can't see a flaw in the logic. I even like the name - "SheepSat", don't ewe?

>The shepherd gives the sheep commands to image an area, the sheep relay
>the image data to the shepherd, and then the shepherd relays it to the
>ground for image assembly. You still have some issues, like
>constellation spreading, but you are fairly immune to bit errors,
>and "sheepsat" failures. Sheepsats might even be more practical for
>missions like FORTE (a recent sat out of Los Alamos that detects
>nuclear tests), where you want to cover the globe cheaply.

>You just can't beat the REconfigurability of a microprocessor
>though. I guess every tech has its strong points. BEAM's counterpoint
>would be to just build another, since it is (or should be) so cheap.

Well, I already acknowledged that point, but it's exactly that flexibility that can cause it to hang from single-bit errors caused by space radiation. I agree, BEAMtek is a good counterpoint to the traditional, not a replacement.

>And I really want one of those butterflys to put on my windowsill in my
>new office. How hard are those kits to assemble?

Piece of cake, especially if you already know how to solder. Just keep it away from floppy disks (high-intensity magnets and a solar-powered degausser isn't magnetic media friendly!).

-Dave

On 4/29/98 7:47:10 AM, Landon Mills wrote:
>why do you wait till all odd hours to get on??

Because I'm an odd sort of fella... Actually, my day is quite busy tracking inventory, processing orders, throwing frisbees (for the shop dog, "Glue"), and answering telephones. I don't have much help here yet.

>I think we should use beam and Microprocessors in the pathfinder
>style robot

I think it's viable. But let me throw this wrench into the works: In regards to general robotics I don't think legs are the end-all and be-all. I *enjoy* wheeled 'bots. Don't get me wrong - I love what Mark Tilden has done with legged BEAMbugs, but I'm still convinced that a well designed wheeled 'bot will take less time to build and perform just as well as a legged robot built for the same job. Call me a BEAM heretic, but I still feel that the PathFinder/Sojourner layout with wheels was the right way to go.

Debate anybody?

>and wouldn't a walker be better on the rough terrain of the moon
>if you gave it a high step??

But high-stepping requires more time and power to perform. The bogey-system used on Sojourner seemed quite capable. And the "Rambler" series (I think that's the name) of wheeled 'bots built at Sandia labs (they have an excellent website on these) can clear objects equal to their height already. Can't beat rolling for most efficient use of power.

Hey Larry, if you buy and build one of Dave's Magbot Butterflies, let us know how you like it here on Shop Talk

On 4/30/98 1:40:09 AM, Dave Hrynkiw wrote:
>Call me a BEAM heretic, but I

BEAM heretic.

>still feel that the PathFinder/Sojourner layout with wheels was
>the right way to go.
>
>Debate anybody?

Well Sojourner pretty much had wheels on legs if you think about it, didn't it? I'd call that a compromise.

On 4/30/98 2:23:49 AM, Gareth Branwyn wrote:
>Hey Larry, if you buy and build one of Dave's MagbotButterflies,
>let us know how you like it here on Shop Talk

I'm seriously considering one of the PhotoPoppers, too. Should make a fine companion to the laser pointer for messing with the cat.

Tim,

I've forwarded your messages on the Idaho Humane Society, they've assured me that they already have an ongoing "interest" in your practices.

I'm working on that thatt that stutter.

Well, this very merry and active chat has proven to me that I don't know fuck-all about what any of you are talking about. :-) Microcore, nervous nets, frobwitzes, flux capacitors...I feel like one of those housewives our circulation manager just hired to build ads for the newspaper as they try to figure out QuarkXpress...

If it makes any difference, I have a bona fide general interest in robotics, but absolutely no practical experience with either robotics or electrical engineering.

I checked the various hyperlinks that emerged from the sage keyboard of Gareth at the start of this thread and I'm still lost in space.

Can anyone suggest to me a good resource to get started with the basic mechanics and electronics involved in robotics, and perhaps a reference to a kit developer oriented towards complete newbies?

Peter, I think the Beam Tek site (esp. the FAQ and glossary) and Dave's Solarbotics site (URLs above) are good places to start.

Dave's kits are really great and are highly recommended. They are very professionally put together (good docs!) and cheap for the cool stuff you get. I got my first one (the Solarspeeder 1.1) and am anxious to put it together (in my voluminous spare time). The little pager motor is so cool! The kit looks very straight-forward and relatively easy (if you have basic soldering skills). And if you screw it up, you can send it back to Dave and he'll try to get it working.

The only other BEAM kit that I've seen is the CYBUG. (Dave, are there other kit makers besides you and Craig Maynard?)

For a CPU-oriented approach to small autonomous bots, there's the MIT Mobile Robots book. It covers everything from basic soldering and construction tips to sensors, software, etc. The book mainly focuses on one bot: the Rug Warrior which you can also get as a kit for *only* $600! It uses the 68HC11 CPU which I've always dreamt of playing with.

There's also a new bot book out called Practical Robotics by Bill Davies. I haven't seen it yet, but it's billed as a good book for beginners. Maybe it's time to phone the library.

Fortuitous timing: There's a piece on CNN Online today about BEAM's "big god" Mark Tilden. There's a short piece, some video, audio and pictures of his bots. Man, this guy really puts the "A" (aesthetics) in BEAM--gorgeous work.

On 4/30/98 3:33:07 PM, Gareth Branwyn wrote:

>The only other BEAM kit that I've seen is the CYBUG. (Dave, are there
>other kit makers besides you and Craig Maynard?)

There's quite a few, but there are quite a few "me too" type manufacturers. One other interesting company is Lynxmotion - they've got some pretty nice stuff too.

On 4/30/98 3:33:07 PM, Gareth Branwyn wrote:

>For a CPU-oriented approach to small autonomous bots, there's the MIT
>Mobile >Robots book. It covers everything from basic soldering and
>construction tips to sensors, software, etc. The book mainly focuses
>on one bot: the Rug >Warrior which you can also get as a kit for *only*
>$600! It uses the 68HC11 >CPU which I've always dreamt of playing with.

Mmmm 68HC11... I play with 'em just about every day - Pegasus has got about a zillion of 'em. The Flight Computer & INS have got real CPUs (3x68030 variant & i860(?) respectively), but just about every specialty box (Pyro drivers, multiplexers, Thruster drivers, ...) is run by a 68HC11.

During my MIT days, I entered the 6.270 contest, where we got a kit of LEGOs, sensors, & a 68CH11 controller board. The idea was to build a 'bot that would push pingpong balls to the opponents side. 'Twas fun, but we didn't do very well.

Oh yeah...I have the 6.270 Robot Builder's Guide around here somewhere. So what was your bot like, Larry?

And...if you ever wanna slip a 68HC11 out of the stock room and secret it off to your ol' pal Gar...

Well, it wasn't a very good robot - it beat the placebo (a cardboard box) but that's about it. The field was hill-shaped, sloping up toward the middle, and one side was black & the other white. We had tilt sensors & leds/photoresistors so we could determine our location. All we really attempted to do was to sweep the field back & forth in a pattern & then push the balls to the other side. We got bogged down in making our sensors work & interpreting the data. I think some BEAM design principles would have helped out.

We put together our boxes in Chandler, so I don't think I have access to any raw 'HC11s. Sorry Gar...

On 4/30/98 3:33:07 PM, Gareth Branwyn wrote:
>It uses the 68HC11 CPU which I've always dreamt of playing with.

Your dreams are much more disturbing than mine, Gareth. You are now officially scarier than Marylin Manson. :) Of course, I think he's about as scary as Count Floyd...

>Your dreams are much more disturbing than mine, Gareth. You are now
>officially scarier than Marylin Manson. :)

OK, so maybe "dream" is kind of an exaggeration. I've been interesting in microcontrollers for years but have never had an opportunity to play with 'em. I always read about the 68HC11 and the BASIC Stamp in various books and articles about robots and embedded systems. It's all slightly outside my league (I know next to zilch about programming), but I'd love to learn (again, in my voluminous spare time).

Dave, here's a few follow-up questions:

Re: Mark Tilden's ROSWELL bot
Are there any images of this on the Net that you're aware of?

On Tuesday, April 28, 1998 04:01 PM you wrote:
>Mark's also working on a new series of solar-powered BEAMant Photovores
>that are so active and animated, you would think they're battery
>operated. These are truly slick little machines, and since other
>robots this size (3" square, 3/4" high) usually have dead batteries
>in 20 minutes, we're looking into marketing these as experimental
>platforms for hive behaviour for research institutes. Simply add
>light, and watch'em go...for days at a time. Try analyzing hive
>behaviour when you have to keep replacing batteries.

Has there been any research into hive behavior done using BEAM bots?

Or flocking behavior?

On Wednesday, April 29, 1998 01:11 AM you wrote:
>Well, it _is_ a fancy term because there really isn't any other term
>suitable to describe what's going on with these analog circuits. On the
>surface it looks like a simple little chain of inverters, but the
>subtlies of it are far more complex. They don't have theoretical
>physicists at LANL looking at these thing just because of the pretty
blinky lights!

Can you tell us more about LANL's research into this?

And, can you tell us a little about the BEAM games. How often do they happen? What sorts of events? When's the next one?

And, since this is the last day of the chat, can you also give us some thoughts on where you see BEAM possibly being in the next 5 years? 10 years? And what do you see as some of the likely future applications?

Thanks so much for stopping by this week. It was a pleasure having you here. Good luck with Solarbotics and your own BEAM projects.

BEAM talk participants: We're going to be doing a drawing for a BEAM kit for topic participants, so stay

Curious, will the topic simply disappear from the list, or does this simply mean that the important guys aren't officially part of this???

Oh well, back to the BEAM list! (:

Richard Piotter

On 5/3/98 4:39:19 PM, Gareth Branwyn wrote:
>And, since this is the last day of the chat, can you also give us
>some thoughts on where you see BEAM possibly being in the next 5
>years? 10 years? And what do you see as some of the likely future
>applications?

In the next decade, I'm expecting to see BEAM technology doing the simpler things that you want done effectively and cheaply. Solar-powered lawn trimmers; window cleaners; data-recorders; space exploration walkers; TOYS. I think BEAM will hit it very big as toys. There's simply something appealing about no batteries and looong life for executive office toys. I'm no visionary, but these are the areas that I think BEAM can be most easily introduced into the "real" world.

>Thanks so much for stopping by this week. It was a pleasure having
>you here. Good luck with Solarbotics and your own BEAM projects.

Thenkyoubeddymuch. 'Twas plenty fun.

On 5/2/98 4:05:53 PM, Gareth Branwyn wrote:
>Dave, here's a few follow-up questions:
>
>Re: Mark Tilden's ROSWELL bot
>Are there any images of this on the Net that you're aware of?

No, not yet. That'll be part of our major website update. We're actually hiring somebody explicitly for this purpose in the next month or so.

>>at a time. Try analyzing hive behaviour when you have to keep
>>replacing batteries.
>
>Has there been any research into hive behavior done using BEAM bots?
>Or flocking behavior?

Yes, Mark has some pretty cool little tilebots that self-assemble into a large "plate" of bots. Not sure of the final concept, but looked pretty! ;>

But there hasn't been any sincere effort put into it using BEAM robots....yet....

>>On Wednesday, April 29, 1998 01:11 AM you wrote:
>>They don't have theoretical physicists at LANL looking at thing
>>these just because of the pretty blinky lights!
>
>Can you tell us more about LANL's research into this?

Well, some of the technical papers available at the LANL site describe some of this. There hasn't been too much published, but Brossl Hasslacher (the guy holding Walkman in the "Robots Rising" special) is the man behind the theory. Above my head.

>And, can you tell us a little about the BEAM games. How often do
>they happen? What sorts of events? When's the next one?

The BEAM games are an annual event, now in it's...5? 6th year? Something like that. The last one was in Lucknow India, and was a superb event. The hosts did an amazing job of arranging the event and accommodating all of us. They've requested to do it again this year, and it's scheduled for this Mid-November. Sorry, the exact date escapes me, but there are links to the India games that'll give you full details if you want to go. They've got a great deal going too - if you get yourself to India, they'll put you up and feed you for the duration of the event. Very nice facility too (for India, that is).

As for the events, there's quite the variety - there should be something for everybody. Solaroller (solar drag racing), Photovore (light-eating robots), Aquabots (in the outdoor pool!), and Walker robots. There's many events - again, look at the event website, and it'll have the full information online. If you have a robot, there'll be a category for it to compete in.

Richard, this topic is not going anywhere, but Dave likely is. Of course, he's always welcome here, but he was only officially on-board for the past week as our latest "Geek-in-Residence."

I'm certainly going to continue exploring BEAM (as a novice bot hobbyist and tech journalist) and will post my explorations here. We'd love to hear about other's projects.

What BEAM stuff have you done,

From: Richard Piotter (richfile@prairie.lakes.com)
Date: Monday, May 04, 1998 10:06 AM

>Richard, this topic is not going anywhere, but Dave likely is. Of
>course, he's always welcome here, but he was only officially on-board
>for the past week as our latest "Geek-in-Residence."

Cool!

>I'm certainly going to continue exploring BEAM (as a novice bot
>hobbyist and tech journalist) and will post my explorations here.
>We'd love to hear about other's projects.
>
>What BEAM stuff have you done,

Well, I've found that I'm good at the mechanical aspects. I've salvaged capstan flywheels from camcorders and turned them into wheels, complete with ball bearings! I spun the wheel and it went over a minute off the ground!

I built a tiny solarroller with a microcassete rubber pressure wheel and ran a set of slot car wheels through the pivot hole. I made bushings for it from a yoke plug in a monitor. Spins nice, cause it only has the edges touching.

I've also got a walker frame, but I damaged one of the gear boxes recently. I think I can get away with turning it around though, since the teeth are only on one side.

The problem is I've never got any of the electronics to work. I have some 1381's in the mail, but they haven't arrived yet.

I've had no luck with the FLED SE, and I haven't had a chance to try any other of the more complex SE's. I've always found myself building the stuff, but never adding the electronics! (: I have a model railroad that looks awesome, and I have so many buildings They all don't fit on it, but the train doesn't even run!!! (:

It would be interesting to try to make a BEAM model railroad! It would get power, of course, but maybe have it sense if other trains are near by be measuring current draw or something. Could be a fun project. Don't know if it'd become BEAM, but would be interesting!

I got a microcore on bread board, but it won't even come on. The bread board I'm using is a piece of junk and one wire is magnet wire cause I don't have any thing decent that long. I'll fix it up later. I threw it together in a few minutes and never got it to run. I'll figure it out later when I got time.

Richard Piotter