Swarming

bxgirten

This post probably belongs in the "Robotics" section but since I deal with the Javelin Stamp, I thought I'd try here first.

I want to have my BOEBot respond when it is being "summoned". The thought is: I would have a "Queen Bee" bot summon the swarm to collect within a short distance of itself. I don't want to use radio control - I just want the swarm of bots to collect in my backyard (or whereever) when they are notified.

Is this a job for IR?

Thanks in advance.

Beau Schwabe

At first I would think ultrasound might work better than IR. What kind of distance are we talking that the "drone" bots might "bee" away?

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Beau Schwabe

IC Layout Engineer
Parallax, Inc.

allanlane5

IR's major benefit is that it is cheap. Its major drawback is that multiple senders can easily jam each other. Its other drawback is it can be distance limited (around 50 feet?) -- depending on how you rig the transmitting IR-LED. It can also 'bounce' off white-painted surfaces.

How about two recievers -- one shrouded (with metal or paper) to be 'directional', and one open to be omni-directional. This means 'calls' have to be repeated a few times so you can check both sensors. You can then use the shrouded one to find the direction of the 'queen'.

Ken Gracey

bxgirten,

Like Allan said. . .

One useful aspect of infrared is that a robot with a shielded detector could receive IR signals from a transmitting robot when they're facing eachother. This helps a robot determine whether the other "object" it sees is a wall or a robot. Therefore, to do this right, you need both infrared and ultrasonic. Additionally, you need to use a digital compass. Open-loop roaming is quite limiting - your robots need to gather in a predominant location while avoiding objects and walls. I think the best approach is with three sensors: infrared, ultrasonic and a compass.

I'm curious about your outcome because I'm working towards the same with my Penguin robots. I've got the three sensors I described on the Penguin's PCB. But these are walkers. . . really tough to program. Okay, I'll stop the Penguin discussion - others are going to get bored of my fascination.

Ken Gracey
Parallax, Inc.

bxgirten

Guys,

Thanks for the great replies!

Beau, the distance from the summoning bot could be up to 300 feet and could move around in a field.

I never thought of ultrasound. How would that work? Would an ultrasonic receiver hone in on a particular frequency?

Allan, good point about the jamming. Plus the IR would probably not work if an object (or objects) were in between the summoner and the summoned - let alone blades of grass.

Ken, is the compass used to sample the strength of the ultrasonic signal? As humans, I know our brain samples a sound and determines which ear heard it first. Where does the digital compass figure in?

This forum rocks!

Ken Gracey

bxgirten,

The digital compass is an aid that would assist the robots in meeting in a predominate location. When I want to be around other people I go to the grocery store, which is in the NE quadrant of my 7.5' USGS topo map. The store is against a mountain, so all I have to do is head NE. Okay, that's a bit further than your robots will travel. Rather than simply "grouping" where other robots are found, you could supplement the behavior by meeting in a specified direction of travel, perhaps until a real obstacle (such as your fence) is encountered.

The compass senses the earth's magnetic field and provides 127 different headings for your robot (the HM55B).

Since you want to do this outside in a large area you might want to consider using GPS. To see how we use GPS on our model airplanes download the video on Parallax.com / Resources / Videos / Customer Applications. You'll see that it also provides plenty of resolution for what we are doing.

Then again, I don't want to complicate a project that had a vision and a reasonable approach. Whatever you do, start simple.

- Ken

- Ken

Ryan Clarke

Couldn't you just use the 'summoning bot' as a beacon? Then you woudn't need the two-way communication (like with the IR)- then it wouldn't matter if that thing in front of you was another bot or a wall, your response could be the same...

or am I over simplifying this?

Ryan

Paul Baker

Something which hasn't been discussed yet which should be mentioned is, having your bots in an open (sunlit) space will have a deleterious effect on the operation of your IR system. IRs are pulsed to help combat this, but as the transmitter and receiver are seperated the Signal to Noise Ratio becomes worse to the point where the pulsed IR becomes only a small fluctution of the ambient IR levels. In direct sun, you'll experience problems detecting the IR after only 20' or so. Ultrasonic suffers similar problems with the signal intensity decreasing at a rate proportional to 1/r^2, getting a detectable signal over a distance >100' becomes difficult (using easily availible ultrasonic sensors). I don't know of a non-RF communication scheme capable of satisfying all your requirements.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔

bxgirten

Maybe what I need here is a system of relayed beacons? Queen bot sends an ultrasonic signal to a worker bot within 100'. That worker bot delays a few milliseconds and sends an ultrasonic signal to other worker bots.

Sounds good in-theory.