Simulation of Intelligent Robotic Colony

Rxke · 2004-08-22 15:37:04

About triangulation...

i was thinking (for a Mars mission, heehee...)

one base station, after landing it catapults away some distance (preferrably outside operation radius of bots, centre i base station) 4 or more independant beacons that work on solar-cells, emitting coded bursts, but... you should be able to tell the bots (a swarm that came with the lander) where the beacons are. Probably possible, if the base station has some kind of narrow-beam antenna to check related placement of the beacons after deployment, and upload that info to the bots before their deployment.

So the bots have a rough idea of direction, using matrix calculation of signals
(too tired, probably utter nonsense...)

Rxke · 2004-08-22 15:38:47

Grypd, in Belgium the parking meters work on solar cells... Now you're giving me an idea!

Rxke · 2004-08-22 15:50:25

Ugh. No, wait...

Basestation calculates XY coordinates' "radio-fingerprint" through triangulation and tells bots "try to find that place" Bots will seach least resistance route to that place.

Do I make sense?

I mean... base knows where beacons are... and could calculate for a given place in a XY plane how those beacons would 'smell' (how do you call that?) if you looked from that point to base.

say place XY gives .3 beacon1; .5 beacon 2; .77b3 etc.
Al you have to do is fit a directional antenna locked towards base, and ...

Hmmm... I see snags. Nevermind.

Grypd · 2004-08-22 15:56:21

You have forgotten about GPS.

We will have this if only so as to know where our poor astronauts are and to provide communications. That will give you more reference points

Rxke · 2004-08-22 16:11:06

Yes, but GPS needs a dedicated set of sats... Will be awhile and lots of $$$ before that is in place...

Grypd · 2004-08-22 16:45:54

Ah but to provide the communication planned for mars it is envisaged that mars will have a form of internet. This needs a set of satelites and these could easil have a GPS function as well on their carrier band. And another option is to provide so called power tower type location radio signals.

these towers would be set at exact distances from each other have a signal from 3 and you can triangulate

John Creighton · 2004-08-23 10:29:30

Some BEAMbot LInks:
This links give alot of how to stuff to get somone started. I'll look around to find the design I thought was interesting.

http://library.solarbotics.net/]Conan the Librarian welcomes you to the BEAM Reference Library

http://www.talkingelectronics.com/Proje … tml]ROBOTS,
ROBOTICS and AUTOMATION

http://faq.solarbotics.net/FAQ.html]Here is an FAQ

John Creighton · 2004-08-23 10:36:19

Oh neat this Paper discribes how the actual control of the walk works. I like this BEAMbot in that it is about the simplist walker that has resonable functionality. It is still not the design I saw before that I want to try. However this design only requires two motors so it should be realy easy to build. If I ever start my getentic BEAMbot design program this bot will be one of the orginal parents.

http://www.solarbotics.net/library/pdfl … Controller for a four legged walking machine

The robot only has four gates. A quick forward gate and a slow backward gate. I think if I build this one I will have to figure out how to modify it so it can turn.

John Creighton · 2004-08-23 10:48:36

Ahhh... Here is the one I wanted to try and build

http://faq.solarbotics.net/patent/patent.html]Mark Tildens Nervous Net Patent

This can also be found on thehttp://www.uspto.gov/patft/]US Patent Database.

John Creighton · 2004-08-23 10:58:40

Anyway back to GPS, if I recall correctly each satellite modulates a carrier frequency with a unique pseudo random sequence. Then I think information about the time and location of the satellite is then modulated onto of the modulated signal. The receiver must internally generate this signal and based on how much the signal must be delayed to get lock on the receiver knows how far away the satellite is. This system could be used for both ground and satellite beacons. For a robot to use this information it will probably need a correlater or some system that incorporates a phase lock loop. The advantage of a ground beacon is it would not have to transmit time and position since it does not move.

John Creighton · 2004-08-23 14:11:36

After reading the paper:

http://www.solarbotics.net/library/pdfl … Controller for a four legged walking machine

and thinking about it some I thought of a way that a two motor BEAMbot could walk in a circle. Recall that this BEAMbot walks forward by leaning to the left with its back legs then pulling its self forward with its left front leg. Then the robot leans to the right and pulls itself forward with its right front leg. If the robot could pull one of it front legs back well both front legs are still in the air and pull the other front leg back well it is still on the ground then the BEAMbot could walk in a circle.

John Creighton · 2004-08-24 10:45:16

More on:
http://www.solarbotics.net/library/pdfl … Controller for a four legged walking machine

After reading more of the article I found that this BEAMbot walked without lifting its feet off the ground. This could probably be changed by added a power amplifier to the motor. Or perhaps a simple CMOS switch that can work at higher currents.

I also noticed that this simple BEAMbot by letting pulses propagate in the ring. When the BEAMbot is walking forward one pulse propagates and when it is walking backward two pulses prorogate. I wonder if the ring was lager then four nodes if there would be any problem of pulses running into each other.

If there would be then why not use a shift register instead of nervous net. Another loop type architecture that would allow pulses to travel in a ring is a transmission line. Simulated transmission lines can be built with capacitors inductors and restores. These principles are sometimes used to build wide bandwidth amplifiers by replacing the inductors by transistors. With the right nonlinear gains stable pulses could propagate around the ring. Another thought I had is replace one of the nodes is the nervous net with a flip flop and have a switch where the switch, switches between closing the loop and clocking in new data. When the loop is closed this flipflop could help insure that the pulses in the ring don’t run into each other.

About the walking in the circle, if more time is needed to move a front leg back well the font legs are in the air some kind of counter weight could be placed on the rear of the BEAMbot. This would mean it would take a longer time for the feet to come back in contact with the gound.

Rxke · 2004-08-24 11:23:17

John, do you know http://www.spiderland.org/breve/]breve ?

I've got to admit I'm totally awestruck by it, because of its "Walker" implementation, which I first saw, yeaaaaaars ago, demonstrated on a biiiiiiiig uni computer.

It uses n IMO quite simple to learn language, and has again IMO quite some capabilities.

And it's free!

"Study Artificial Life & Decentralized Systems
breve is a free software package which makes it easy to build 3D simulations of decentralized systems and artificial life. Users define the behaviors of agents in a 3D world and observe how they interact. breve includes physical simulation and collision detection so you can simulate realistic creatures, and an OpenGL display engine so you can visualize your simulated worlds."

I've been tinkering a bit and it's great. Also just found a support for neural-net implementations... The list seems endless.

John Creighton · 2004-08-24 11:26:43

John, do you know breve ?
I've got to admit I'm totally awestruck by it, because of its "Walker" implementation, which I first saw, yeaaaaaars ago, demonstrated on a biiiiiiiig uni computer.

Rxke, that sounds great! I will have to learn it. Maybe we can network a bunch of these simulations together using our spare CPU cycles. Of course we still have to define our goals.

BTW anyone good at databases. We should create and link to several databases of designs and components.

Rxke · 2004-08-24 12:45:51

The mailer list seems oddly quiet, for a superb project like that... I have the impresion quite a lot of people use it in academics, but most of them don't seem very eager to share their results, exept for some nice screenshots, pity....

(to be continued shortly, got more comments)

John Creighton · 2004-08-24 13:09:29

What about communicating between the robots and computers. I know on earth that http://electronics.howstuffworks.com/bl … ]bluetooth is a widely used standard for wireless devices. If it is a good standard why reinvent the wheel. I am not sure what it entails. I will have to do some reading.

Rxke · 2004-08-24 13:15:01

OK.... Be warned: Rxke's Krazy Plan part 3.5... ramblings...

http://www.newmars.com/forums/viewtopic.php?t=1925]In this topic, I half explained what I wanted to do, using mainly open-source, cheap hardware (ie hundreds of bots) that people can buy (but not necc. lay their hands on, say it will be engraved w your name or so... More like sponsoring) for a serious, but still affordable amount of money, those bots get sent to Mars, paying back lots of your initial outlay... The idea was, if you can sell bots to sponsors that cost, say 2000$ apiece, but weigh only 200 grammes, you get an almost free ticket to orbit.

Basic idea: send a lander with a furnace, and a lot of robots that fill and empty furnace with Si to make solar panels. If this works, you have a powerplant on Mars (robs can clean panels etc too blahblahblahhhhh)
Follow-up missions could plug in to this plant and add to its capabilities. Slowly, incrementally building real hardware... From the land.

I've got more (probably wildly delusional) details in a folder on my 'puter under the caption RXKE'S KRAZY PLAN, so I don't take it too serious, but... I think a form of 'distributed' designing and building *is* possible, and *can* be cheaper, if you *insist* on it being cheaper. A lot of volunteerwork etc... A lot of time wasting squabbling, too...

Sooo.... where does breve come into this?

It might be the start of everything: set up an open-source project (like the Mars Simulation mission) using breve...
(Ill call it RCP4M for now, Rxke's krazy plan for Mars, heehee...)

Everyone can start up such a project, using for example http://sourceforge.net/docman/display_d … ourceforge to host the code and give other people acces to enhange it...

which answers

BTW anyone good at databases. We should create and link to several databases of designs and components.

for a large part...

Say you start with the idea: make a surfacemap of a part of Mars, built a lander (As design is of course virtually non-existant at this point, just an impressive looking thing with some ramps, to let the robs in and out...)
And give some extra proto-robots. people can download it, run it in the background, upload results etc... Coders can make better bots, etc etc...

It is something the Society could do, or give a little backing, even if it was only an occasional mention.

Rxke · 2004-08-24 13:19:23

Bluetooth is ok insofar that it is not power-hungry, but it's fairly limited in distance. Of course, how far do one wants to go?....
You could deploy repeater beacons or something. KISS instead of megazillions overkill GPS. Come to think of it, robs could be repeaters themselves, maybe... Hmmmm....

(Of course, a GPS-like mast-antennae system is fairly KISS too, so what gives most bang for the bucks... )

John Creighton · 2004-08-24 13:24:30

OK.... Be warned: Rxke's Krazy Plan part 3.5... ramblings...

I think this is something the mars society or another private organization could do. Lots of simulations and tests would have to be run first to demonstrate a reliable design. The mission wouldn't have to be fully self sufficient. It could instead be a technology demonstration mission. Some science could even be done on it. One of the first steps has to be what are the objectives and how should we rank one objective with respect to another. Once the objectives are known some computer algorithm (probably a genetic algorithm could go about and optimize the plan). I think tough that the private mars mission should only be one of the niches in the ring of simulations. Each ring would have a different set of objectives and weights. Each ring would share designs from the other rings.

John Creighton · 2004-08-24 13:26:55

Come to think of it, robs could be repeaters themselves, maybe... Hmmmm....

robonet

SpaceNut · 2004-08-24 13:33:36

Sort of like a roaming cell phone tower.

Rxke · 2004-08-24 13:49:55

Yes, cellular (GSM in Europe) have a capability to send their coordinates, but it's fairly crude, mawbe because the masts are spead out far from eachother.

John, I *know* it all sounds crazy (Krazy) rambling, I should take some time to pour it into a structured cadre... so it makes more sense. Your 'rings' are implied in the krazy plan, i just ommitted mentioning it...

lots of simultaneous developing of ideas, filling in blanks, setting up an organigram(sp?)...

But I'd really love to have a proof of concept of an 'ant-robot' operated manufacturingplant on Mars as the outcome...

Dreams dreams dreams...

John Creighton · 2004-08-24 14:09:25

Another thought. CNC Machines are really big. I wonder how far image processing is where an ant like robot could chisel away at a part. The robot ants could have tools that attach to them like drills and sanders. Another ant could collect all the fillings and return them to the furnace. Granted if you really need precession than a CNC machine will be needed. BTW the image processing software dosen’t needs to be on the ant. It can be on earth or on local computers wherever the antbots operate.

After a little more though. If the ant just secures the part down it doesn't really need much image processing. It just has to know what coordinates to drill at. BTW I was thinking of ultrasonic imaging. This kind of imaging has the advantage that you can see at a smaller detail then you can with visible light.

Grypd · 2004-08-24 18:03:46

Cellular phones in Europe operate by being in a triangle of phone masts this gives a possible location and multiple channels of communication delivery. This method can give you a means to communicate with a large number of Robots. The emergency services in Britain use something similar and up in the highlands it has been modified to provide for location finding.

If a large numbers of Robots are used you can use the cluster type of Robot where a large number of robots can follow one leader type. The commands can be given to the leader but if it was to be put out of communication it would then be passed to another robot in the chain. This way a large number can be used and a lot done without the use of too many hard to get channels. If we manage to get a half decent AI or command computer then it can be the one in charge of these swarm bots.

John Creighton · 2004-08-28 15:19:05

I started a discussion about BEAM technology at:

From the replies it appears I had some false assumptions:

http://groups.google.ca/groups?hl=en&lr … mystifying BEAM technology; Implicit Control; ...

New Mars Forums

Announcement

#26 2004-08-22 15:37:04

Re: Simulation of Intelligent Robotic Colony

#27 2004-08-22 15:38:47

Re: Simulation of Intelligent Robotic Colony

#28 2004-08-22 15:50:25

Re: Simulation of Intelligent Robotic Colony

#29 2004-08-22 15:56:21

Re: Simulation of Intelligent Robotic Colony

#30 2004-08-22 16:11:06

Re: Simulation of Intelligent Robotic Colony

#31 2004-08-22 16:45:54

Re: Simulation of Intelligent Robotic Colony

#32 2004-08-23 10:29:30

Re: Simulation of Intelligent Robotic Colony

#33 2004-08-23 10:36:19

Re: Simulation of Intelligent Robotic Colony

#34 2004-08-23 10:48:36

Re: Simulation of Intelligent Robotic Colony

#35 2004-08-23 10:58:40

Re: Simulation of Intelligent Robotic Colony

#36 2004-08-23 14:11:36

Re: Simulation of Intelligent Robotic Colony

#37 2004-08-24 10:45:16

Re: Simulation of Intelligent Robotic Colony

#38 2004-08-24 11:23:17

Re: Simulation of Intelligent Robotic Colony

#39 2004-08-24 11:26:43

Re: Simulation of Intelligent Robotic Colony

#40 2004-08-24 12:45:51

Re: Simulation of Intelligent Robotic Colony

#41 2004-08-24 13:09:29

Re: Simulation of Intelligent Robotic Colony

#42 2004-08-24 13:15:01

Re: Simulation of Intelligent Robotic Colony

#43 2004-08-24 13:19:23

Re: Simulation of Intelligent Robotic Colony

#44 2004-08-24 13:24:30

Re: Simulation of Intelligent Robotic Colony

#45 2004-08-24 13:26:55

Re: Simulation of Intelligent Robotic Colony

#46 2004-08-24 13:33:36

Re: Simulation of Intelligent Robotic Colony

#47 2004-08-24 13:49:55

Re: Simulation of Intelligent Robotic Colony

#48 2004-08-24 14:09:25

Re: Simulation of Intelligent Robotic Colony

#49 2004-08-24 18:03:46

Re: Simulation of Intelligent Robotic Colony

#50 2004-08-28 15:19:05

Re: Simulation of Intelligent Robotic Colony

Board footer