UPDATE: More info on the subject here: https://www.robotshop.com/letsmakerobots/node/5194
Original post:
I have this thing in the back of my mind that "the next step" for me must be my robots talking to each other.
In the very little time I have for building & playing & researching these days, I have been playing arouund with pulsating lights, robots feeling the bass from a stereo, the very useful "external timer" commands within the Picaxe libraries and such.. But have not found that killer app that I was looking for.
Staring out in blank nothing, thinking.. oh, if only I could buy like 10+ Easy Radio transmitters.. but that would be quite expensive.. and so I just browsed, and wow!
You can really get some dirt-cheap wireless communication devices now! Holy cow, tomorrow I will browse more and order some.
You could also use IR You could also use IR sensors that cost less than a buck for 2. Of course you can only send numbers from 0 to 255, but if each bot knows what the number means then it works. You also have to worry somewhat about line of sight but IR bounces a lot. I got an IR sensor to work when the sending and receiver had a kitchen table in the way
Problem is both line of seight, and the processing they take: With an external reciever - like the ones I link to above - you can background-task the listening. The Picaxe has interrupts, check the setint command.
And then all the sunlight-flourecent-light-problems, I’m not too happy about them
The cheap way toys and such communicate is usually with IR. The Wowee robosapiens, quads, what have you react a little to each other, through their flashy IRs blinking at each other. Cheap and works, but think they are typically around a foot (30 cm) from each other when "yakking".
A clue from one swarm robot project I sorta remember seeing, is to have each bot be a node. One robot gets a message, reacts to it, and passes it on, repeating the message. So you have say 3 robots, 1 in a hall, one at a corner, and another around the corner, the hall-bot generates a message, which the corner bot picks up, and begins moving to, and it reflashes the message to the hidden-botwhihc gets it and travels futher without having line of sight to the hall-bot. Termination could be after the message is reacted to, to sent back an acknoledgment, so the hall-commander-bot can think of something else the underling bots can react to.
But cheap RF is hard to beat for getting more faster better information out and back.
I was looking at your $5 Rx/Tx units and it seems that they are not addressable. Not that this is a huge deal, but at a minium you will always have to send 2 numbers --the first sets the "address" while the second is the data-packet.
The sun radiates IR, the same goes for candles and about 50% of all my inhouse lighting. So critical messages can be lost when using IR. Radio is much safer.
I have been working on doctoring and porting a speech engine written in qbasic4.5 not only giving it access to my own AI but also stubbing in some adanced calculation ability’s and plotting along with access to dos and there by access to programs and the web.
the point is if it’s speech engine and AI alone can be ported into other basics like Parallax or AVR,axiom then we could use it as a means of bot to bot comunication. it uses phenoms to pronounce words and it could also sample sound for phenom recognition. Just some ideas…
Those are indeed dirt cheap! And a lot cheaper than using bluetooth as I was considering (to establish a wireless link between my PC and an Arduino).
However I read somewhere that they’re not so stable if you wanna send/receive larger amounts of data, so one should use something called the manhattan method for data reconstruction…Anyone has experience with this they would like to share?
This “warm-up” method was invented by Frits and works flawlessly. I use it on Walter for RC and I have also used it to send data back and forth from my laptop (using processing) to my picaxe chips. I have sucessfully driven my robot using a virtural joystick on the screen and sending serial data via RC. I average about 98 or 99% clean data transfer with only a little “cough” from time to time.
Thanks I’ll be reading up on that… As well as the manhattan method. Even a 1-2% dataloss may cause severe problems for me if I don’t take the proper precautions.
