Simulation of Intelligent Robotic Colony

John Creighton · 2004-08-28 15:37:48

Hmmmm, interesting a BEAM discussion group.

http://groups.yahoo.com/group/Beam/]htt … roup/Beam/

Grypd · 2004-08-28 15:44:54

Thanks John very interesting site, Maybe give me ideas to sort out all my little bot hiccups.

Rxke · 2004-08-29 02:14:36

Hmmm... wish I had time to read it... next month, hopefully.

Thank you John!

dicktice · 2004-08-29 12:08:14

Shouldn't we be more knowledgeable about how e.g. ant colonies work without recourse to wireless electromagnetic communication? Millions of years behind how they manage to accomplish what they do shouldn't be cast aside, when our design-lives are so relatively short.

Rxke · 2004-08-29 12:23:47

Welll... Ants work with smell and 'sign-language', two things a computer and thus robot is currently not very good at...
Though you're right, and some are working with visual communication 'signlanguage' a little swarm of bots, comm with flahing LEDs... Smell however...

John Creighton · 2004-08-30 09:06:05

Welll... Ants work with smell and 'sign-language', two things a computer and thus robot is currently not very good at...
Though you're right, and some are working with visual communication 'signlanguage' a little swarm of bots, comm with flahing LEDs... Smell however...

If you are going to flash an LED use an IR LED

http://www.solarbotics.net/library/circ … ox.html#ir

http://www.xs4all.nl/~sbolt/e-spider_IR … umper.html

Rxke · 2004-08-30 10:11:11

if looking for IR LEDs, go around carboot sales etc, and pick up the older mono wireless headsets: plenty of LEDs in their base-stations, and receiver in the 'phones... Also broken Handycams with 'nightshot' have them (and lots of interesting motorstuff, gears etc...)
Oh yes: all kinds of remotes too, of course.

Rxke · 2004-09-03 03:31:55

http://fab.cba.mit.edu/]Fab Lab....

John Creighton · 2004-09-03 07:44:47

That is a pretty cool link Rxke. BTW can you do injection molding at home?

Rxke · 2004-09-03 10:37:57

Yes, I thought so, too... Dig a bit into the site, lots of nice stuff...

You mean me able of doing injecion-molding at home?
If so: No, I'm more into small-scale soldering stuff etc...

Sorry for the telex answers, but still quite busy...

Rxke · 2004-09-03 10:43:05

http://www.seattlerobotics.org/encoder/ … ]Injection molding DIY

Extremely Quick Google brought up this... Coincidentally, it's a guy who's making robots(!)

Looks less complicated than I thought... And a lot less expensive! I really thought this was all but unfinanciable (spelling????) for private persons...

John Creighton · 2004-09-03 11:40:51

It looks like he machined the mold. I wonder if you could do something simple to make the mold like clay or sand casting.

John Creighton · 2004-09-03 11:42:55

That link you gave Rxke is really good. It is too bad I don't have more time to try some of this stuff out.

Rxke · 2004-09-03 11:45:02

I Know the feeling!

Rxke · 2004-09-06 15:07:07

It looks like he machined the mold. I wonder if you could do something simple to make the mold like clay or sand casting.

What exactly did you have in mind?

If a Mars scenario, then I'd not bother with plastics, per se, metal is 'good' enough: the low gravity makes for very interesting properties: Beams with the strenghth of Iron (because they are made of Fe., heehee) but with the weight of Alu...
And metal hardware stuff (mixes) can be made fairly easy with simple molds, using wax/clay... No high pressure stuff, you could probably make a 'rapid prototype-printer,' working with wax instead of plastics... make the models, encase them in clay, et voila.

OTOH.... Hmmm... clay on Mars??? Hmmm... maybe just use 'Martian Portland cement' instead? (Oh, and recycle the wax, should be quite possible, if heated gently, then filter it etc...)

Rxke · 2004-09-06 15:18:13

Oh, little update: I'm currently a little less busy, so i wanted to dig into breve some more...

And my computer broke down. Working on another one now, but with an older OS which can't run breve. Sigh.

So, off on a reading binge to the links John supplied, the ones about BEAM demystification

Grypd · 2004-09-06 15:55:05

If a Mars scenario, then I'd not bother with plastics, per se, metal is 'good' enough: the low gravity makes for very interesting properties: Beams with the strenghth of Iron (because they are made of Fe., heehee) but with the weight of Alu...
And metal hardware stuff (mixes) can be made fairly easy with simple molds, using wax/clay... No high pressure stuff, you could probably make a 'rapid prototype-printer,' working with wax instead of plastics... make the models, encase them in clay, et voila.
OTOH.... Hmmm... clay on Mars??? Hmmm... maybe just use 'Martian Portland cement' instead? (Oh, and recycle the wax, should be quite possible, if heated gently, then filter it etc...)

Some of the properties of metal under low gravity are still to be fully worked out so you may get stronger materials than that. Also imagine what quality of metal we could get if set in High gravity?

Another material possible to make would be reinforced concrete with the extra support of iron in the mix. This could make for a decent form of shielding.

Rxke · 2004-09-06 16:06:33

Some of the properties of metal under low gravity are still to be fully worked out so you may get stronger materials than that. Also imagine what quality of metal we could get if set in High gravity?

Not sure if I understand what you mean...

I guess lower gravity leads to a different behaviour of molten metal going towards the solid phase... (internal turbulent descending or ascending fluid metal due to temp. differences... Problems phrasin this in English...)
In zero-g there is no such thing, but lower gravity: will it be more or less? Hmmm. Depends on viscosity of the metal I guess... Probably still quite significant internal turbulence, would only stop in near zero-g.

GCNRevenger, you reading this? Any idea how solidifying fluids behave in different gravity enviros? IIRC you studied fluid behaviour?

John: nice discussion going on, understand it subconsiously, most of it, but afraid to get to the bottom of it, it would take some further reading.
Anyway... I'm slowly getting the impression the beam tech is limited... Seems like no-one has been succesful to go further into complexity... Feedbacksystem seems to 'sabotage' itself beyond a certain number of cores.

John Creighton · 2004-09-06 18:34:54

John: nice discussion going on, understand it subconsciously, most of it, but afraid to get to the bottom of it, it would take some further reading.
Anyway... I'm slowly getting the impression the beam tech is limited...

It is my impression that most applications of the technology to date have been rather primitive. I think this is because the majority of academic research in robotics has focused on a microprocessor approach rather then a nervous net approach. The addition of a microprocessor means the robot can be improved after its initial creation by upgrading the software.

A microprocessor is often an easier way to deal with sensors that require precise timing such as an ultrasonic range finder. However today we have programmable hardware such as FPGA’s. Algorithms can be implemented faster in hardware because it is more natural to create a high level of parallel processing. I suggested a single chip that could be used to program a hardware configuration of a Nervous net. It could be based on the same electronic principles of the original nervous nets. Alternatively I think it shouldn’t be too difficult to get the same kind of behavior in an FPGA. I think the cheapest FPGA is about 50 dollars. The once at my school I think are 500 dollars. I think you get some primitive microprocessors for about two dollars new.

Anyway here is my quote:

Similarly programmable circuits could be developed that perform the
same function as BEAM circuits. There would be a write pin, read pin,
input pins and output pins. Complex motions could be described by bit
streams and feedback could delay the propagation of the bit streams.
Learning architectures could be employed to find the optimal bit
streams allowed by the programmable hardware architecture. The end
result would be something that is similar to muscle memory.

John Creighton · 2004-09-06 18:39:31

Seems like no-one has been succesful to go further into complexity... Feedbacksystem seems to 'sabotage' itself beyond a certain number of cores.

Do you have a reference? Anyway, for mono core rings (I hope I got the terminology right) the pulses can be stopped I think by applying a voltage in the right spot of the nervon. See marks Patent of the eight motor spider. Thus if the reluctance alone wasn’t enough to delay the pulses additional feedback techniques should be possible.

John Creighton · 2004-09-06 18:45:29

What exactly did you have in mind?
If a Mars scenario, then I'd not bother with plastics,

Well, I am thinking both about constructing robots at home and on mars. The structural aspects of robotics are just important as the electronics. Anyway, on mars plastics won’t be as good because the UV light will probably break them down over time. However since they can be produced from the atmosphere they will probably require less energy to produce then metal. If you have the material way waste it? An obvious limitation of metal is it doesn’t make a good insulator. It also would produce a lot of friction around the joints. Plastic could be used as wear pads around the joints kind of like cartilage. Plastic could also be used to insulate wires and electronic components. Another alternative might be glass glued concrete.

John Creighton · 2004-09-06 20:14:54

Seems like no-one has been succesful to go further into complexity

Here is a discussion about a hominid toy created from the father of BEAM. It uses a microprocessor but the claim is it still uses alot of beam principles. Be your own judge.
http://groups.google.ca/groups?hl=en&lr … oogle.com] FYI Mark Tilden Signature Series Robosapien

smurf975 · 2004-09-06 22:01:54

I wonder why software simulations can't anwser most questions asked here?

Rxke · 2004-09-07 01:44:16

John, about the complxity 'barriers'... I won't be able to find a reference I'm afraid, for I read about that while on a nightly 'data-binge,' reading stuff, following links, not saving anything... but IIRC it was Tilden himself, in an interview, half complaining they were hitting a barrier (vaguely defined, not like "12 nodes is the max," more like "around 12 you start to see timing irregularities, too big discrepancies etc... " (12 is also a number i just pulled out of my hat, don't remember...)

But maybe that's sufficient (IMO) You can stil digitally (or 'classically' or whatever) 'drive' a robot, made out of gazillions of *separate* smaller beam circuits.
CPU says: "walk" gives power to beam circuit that walks, says "orient to sun", gives power to solarbotcircuit etc...
The digital CPU gives global comments, the beamcircuits 'fill in the details'?

Smurf,

heh, yes. I guess it's because the 'fuzzyness' of the circuitry, the constant fluctuations.. If you run that throug a sim, you'd see your engines run... but apparently erratically, so you won't be much wizer... only if you 'bolt on' a sim of the hardware of the robot (wheels, sensors... ) in relation with its envionment, you'd be able to see whether it behaves 'good' or 'bad'

That's exactly where i think breve could help... it models physical world stuff, and has some neural net capabilities, so...

Rxke · 2004-09-17 02:08:33

http://science.slashdot.org/article.pl? … tid=1]rHex in the news...

link to Slashdot article that mentions PhysOrg article, some good comments with good links.
...and as usual a lot of stupid 'funny' comments, of course...

New Mars Forums

Announcement

#51 2004-08-28 15:37:48

Re: Simulation of Intelligent Robotic Colony

#52 2004-08-28 15:44:54

Re: Simulation of Intelligent Robotic Colony

#53 2004-08-29 02:14:36

Re: Simulation of Intelligent Robotic Colony

#54 2004-08-29 12:08:14

Re: Simulation of Intelligent Robotic Colony

#55 2004-08-29 12:23:47

Re: Simulation of Intelligent Robotic Colony

#56 2004-08-30 09:06:05

Re: Simulation of Intelligent Robotic Colony

#57 2004-08-30 10:11:11

Re: Simulation of Intelligent Robotic Colony

#58 2004-09-03 03:31:55

Re: Simulation of Intelligent Robotic Colony

#59 2004-09-03 07:44:47

Re: Simulation of Intelligent Robotic Colony

#60 2004-09-03 10:37:57

Re: Simulation of Intelligent Robotic Colony

#61 2004-09-03 10:43:05

Re: Simulation of Intelligent Robotic Colony

#62 2004-09-03 11:40:51

Re: Simulation of Intelligent Robotic Colony

#63 2004-09-03 11:42:55

Re: Simulation of Intelligent Robotic Colony

#64 2004-09-03 11:45:02

Re: Simulation of Intelligent Robotic Colony

#65 2004-09-06 15:07:07

Re: Simulation of Intelligent Robotic Colony

#66 2004-09-06 15:18:13

Re: Simulation of Intelligent Robotic Colony

#67 2004-09-06 15:55:05

Re: Simulation of Intelligent Robotic Colony

#68 2004-09-06 16:06:33

Re: Simulation of Intelligent Robotic Colony

#69 2004-09-06 18:34:54

Re: Simulation of Intelligent Robotic Colony

#70 2004-09-06 18:39:31

Re: Simulation of Intelligent Robotic Colony

#71 2004-09-06 18:45:29

Re: Simulation of Intelligent Robotic Colony

#72 2004-09-06 20:14:54

Re: Simulation of Intelligent Robotic Colony

#73 2004-09-06 22:01:54

Re: Simulation of Intelligent Robotic Colony

#74 2004-09-07 01:44:16

Re: Simulation of Intelligent Robotic Colony

#75 2004-09-17 02:08:33

Re: Simulation of Intelligent Robotic Colony

Board footer