Robot corral with a ground-level wire fence - best way?

Ken Gracey

Hey there,

In two months we're headed to the USA Science and Engineering Festival - a super big STEAM exposition launched by President Obama a few years ago. This event brings a mix of 30,000 students, teachers and parents to a three-day exposition. Last time we taught about 1,500 kids how to solder and. . .we won't be doing that again. Didn't do much for business and it was a whole lot of work, too!

This time we're dividing our area up into two big spaces. One will have a 4x8 table with ActivityBots and Boe-Bots, and the related curriculum will be on display. The robots will autonomous so we don't have to mediate fights over RF controls and deal with queues of children who want to play.

But it's the other area of 10x20 that'll be filled with about six Arlo robots that I'm writing this post about. I originally envisioned enclosing the area with a 1' high perimeter "fence" of some kind and use Ping))) sensors as the primary means of avoidance. But I don't want to ship a bunch of materials, set up a fence, see people trip on it, etc. Each Arlo would carry an American flag, too, as you've seen us do in the past in China.

So I'm thinking about a buried wire fence - you know, the technology used in the dog collars. I've searched around on the net a bit for solutions and I'm not coming up with a simple little embedded add-on board. Looks like most people tend to hack the dog collars for this kind of application. And that's fine, but I'd so much rather have a board-level solution.

Before I chase down some hackable dog collars, does anybody have a lead on a small pre-made embedded solution that'll do the job? We don't have time to wire up 555 timers, fabricate circuit boards, etc so it should be mostly off-the-shelf.

Any tips would be appreciated.

Also, it seems like this could be a great add-on sensor for the Parallax line, doesn't it? Seems like a lot of people are building autonomous lawnmowers with the same technology.

Thanks in advance!

Ken Gracey

mindrobots

I'm assuming you can't actually bury the wire in the exposition center's floor, so you'd be taping the wire to the floor to create the perimeter? If that's the case, the why not just use opical sensors and a thick band of contrasting tape? Similar to creating a GIANT Sumo ring. It doesn't introduce a new product/technology into the Parallax line but it is certainly doable in the timeframe with proven technology.

erco

Here's a $41 dog collar/fence system, free ship from Jersey. Only one collar though, I'll sure you'll need many. Might be worth $41 to examine the collar circuit to see how easy it would be to duplicate sans shocking mechanism.

http://www.ebay.com/itm/New-Underground-Electronic-Dog-Fence-Shock-Collar-Electric-Fencing-System-/221287841429?pt=LH_DefaultDomain_0&hash=item3385c87e95

lanternfish

Hi Ken

Have you thought of transmitting an audio tone(s) (AM or FM modulated) through the 'corral' loop and sensing it with the output from a cheapo AM/FM receiver? there may be an small I2C compatible board out there (I haven't had time to look).

Cheers
Lindsay G

erco

mindrobots wrote: »

I'm assuming you can't actually bury the wire in the exposition center's floor, so you'd be taping the wire to the floor to create the perimeter? If that's the case, the why not just use opical sensors and a thick band of contrasting tape? Similar to creating a GIANT Sumo ring. It doesn't introduce a new product/technology into the Parallax line but it is certainly doable in the timeframe with proven technology.

Even better! Of course, you might not know the color of the floor for best contrast (white tape on a dark floor or vice versa) but I had excellent results using chrome Duck tape on white carpet using a Scribbler robot. IIRC you guys used aluminum foil tape at your Expo Line Follower contest, that's probably just as good.

W9GFO

I would think that this would be right up Beau's alley. Yes, it would be a great add on.

Another option: It would take about 300 RFID tags (less than $100 on ebay) to create a sixty foot perimeter.

SRLM

There's an IEEE robotic competition called Natcar that's based on line following. One line is a magnetic field, and IIRC it's not too difficult to come up with a simple analog circuit to detect the field. For this application you'd just want line avoiding instead of following. And here's a page describing the sensing circuit.

It would also be interesting to make the field shape irregular. Then you could do things like have robot-free islands or a viewing peninsula.

xanadu

I thought tape at first too, or magnetic. If there's a crowd of people we gotta entertain them every way possible. You'll get their attention and make them ask questions if it is invisible. If possible an IR system like Roomba uses for barriers seems like a good option.

xanadu

I was also thinking about what could I build to do the same thing. So elaborate idea aside basically.

I'm going to try using bright red EL wire and a CdS cell with blue and green filters. If the filters block enough light, the ambient could be brighter than the EL wire and it should still register, I think.

If the robot had metal wheels you could detect a short between them over something conductive. Or even measure resistance for more options.

I hope that IMUs become good enough for precise indoor navigation soon. This is like duct tape brand band aids for precise indoor navigation.

kwinn

xanadu wrote: »

I thought tape at first too, or magnetic. If there's a crowd of people we gotta entertain them every way possible. You'll get their attention and make them ask questions if it is invisible. If possible an IR system like Roomba uses for barriers seems like a good option.

IR may be the simplest way to go. Maybe 4 corner posts and 2 posts in the middle of the 20 foot run if needed. Place modulated IR emitters at the bottom of the posts to create the perimeter.

SRLM

My main concern with IR is that it won't work if you have people pressing in close to the robot area. I'm imagining a large crowd that wants to watch and are standing shoulder to shoulder. They may block the light from corner posts, resulting in robots that keep going outside the bounds. You'd need a human tape barrier safely outside the robot tape barrier.

W9GFO

The problem with IR is that it will be blocked by people, then the robots will escape. I'm assuming that the corral will be open to onlookers.

xanadu

I just noticed Ken mentioned people possibly tripping over a fence. Sounds like IR wouldn't work there.

I made that hack job RSSI xbee proximity thing, that could keep them in a circle. People walking in between would only weaken the signal making the robots want to stay closer to the center beacon.

kwinn

If a perimeter IR fence will not work the alternative is to flood the area with IR and have the robots stop and turn around when no IR is detected. Each corner IR source sends out a unique code for robots to orient themselves. Something along the lines of how insects are attracted to light.

Adam Wiesler

Maybe you could hack something like this apart--- http://www.harborfreight.com/cable-tracker-94181.html

The wire could be as simple as an extension cord taped down to the floor, and the "sender" clipped onto one end.

dgately

xanadu wrote: »

I made that hack job RSSI xbee proximity thing, that could keep them in a circle. People walking in between would only weaken the signal making the robots want to stay closer to the center beacon.

xanadu,

I like this idea. Did you take this further? Seems like a great way to keep any bots within proximity of a location...

Thanks, I'll be using this method for my own experiments!

dgately

jazzed

Tape one of these (or something like it) to the bottom of each bot and use it for detecting magnetic anomalies.

You could have an army of robot metal detectors. Also, a beach-combing robot would be an essential element for entertainment.

bsnut

Adam Wiesler wrote: »

Maybe you could hack something like this apart--- http://www.harborfreight.com/cable-tracker-94181.html

The wire could be as simple as an extension cord taped down to the floor, and the "sender" clipped onto one end.

This is the best idea and suggestion so far that meets the requirements that Ken is looking for.

Also, this idea or "follow the wire" has been used for years in automated warehouse robots to move pallets for a long time and being put to use in amusement parks for darkrides.

Also, a basic circuit is in the Forrest M. Mims, III book "Getting Started in Electronics" that sold in Radio Shack stores.

xanadu

dgately wrote: »

xanadu,

I like this idea. Did you take this further? Seems like a great way to keep any bots within proximity of a location...

Thanks, I'll be using this method for my own experiments!

dgately

Please do, it really is a viable option that seems to get overlooked. I would love to see someone with better coding skills take a crack at it.

You essentially setup zones.

RSSI > 75 = close to home, use PING to avoid people and roam around
RSSI < 75 and RSSI > 60 = turn routine, recheck RSSI with smaller movements (store RSSI in another variable to compare with last reading)
RSSI < 60 and RSSI > 50 = turn routine, recheck RSSI with even smaller movements (compare with last reading, move accordingly)

I don't think you need a directional antenna. I believe you can also use more than one beacon to compare RSSI with and possibly triangulate. This is why my lack of coding ability ended up another unfinished project.

Publison

How about Magnetic Tape and a Hall Effect Sensor. Will that work?

Phil Pilgrim (PhiPi)

Magnetic tape has very closely-spaced alternating poles -- designed for contact with ferrous metals. As a consequence the field strength at more than miniscule distances is near zero.

-Phil

davejames

Mr. Gracey,

Use one of the cheapy, low-power RF transmitters to broadcast a tone on an unused FM frequency. Ring the robot arena with the antenna. Use one of these to receive the broadcast: http://www.parallax.com/product/27984

...and then use the Propeller to analyze the receiver's output for the correct tone. Detecting maximum amplitude would indicate the fence perimeter.

No tripping, elegant, and it uses a Parallax product to boot!

Ken Gracey

OK, I've gone with the electric dog fence and ordered a three-dog system.

Once it arrives we'll get inside of the doggy collars and report back to this thread.

The infrared, magnetic tape and RFID systems are probably not as well suited for this application due to interference from people (for IR), placement of hundreds of RFID tags, and signal strength with XBee which couldn't maximize our perfectly square area. But these ideas are still appreciated.

So, shall report back!

Ken Gracey

Publison

Phil Pilgrim (PhiPi) wrote: »

Magnetic tape has very closely-spaced alternating poles -- designed for contact with ferrous metals. As a consequence the field strength at more than miniscule distances is near zero.

-Phil

I kinda' though that was the case from reading your post.here:

http://forums.parallax.com/showthread.php/130799-Refrigerator-Magnet-Encoder?highlight=Hall

skylight

There was an experiment conducted on the streets of london by the BBC to promote the Proms I think it was? Anyway they laid a mat down in certain areas and invited people to stand on to the mats upon doing so they heard music out of nowhere, obviously using focussed speakers of some kind, If it could be narrowed enough perhaps the same idea could be used for this perimeter? The added bonus is if people walked into the "Fence" they would be pleasantly surprised to hear some pleasant music....perhaps this?........

http://www.youtube.com/watch?v=us3dQ0nnlHY&feature=kp

Phil Pilgrim (PhiPi)

The Prop makes an adequate RF receiver for this purpose. Here's a schematic:



The attached program is my AM receiver adapted for this app. I cut out a 1.5" x 2" piece of copperclad PCB material, soldered a wire to it, and ran the wire to the solderless breadboard, which also included the 1M resistor. The copperclad was then stuck to the ActivityBot servos with double-stick tape. I loaded the attached program into both the ActivityBot and a QuickStart board. To the IF_PIN on the QS, I connected a long piece of cheap telephone cable. (A single strand of insulated wire will work just as well.) The cable was laid on the floor. When the bot was positioned over the cable the LED would flash, indicating a signal over threshold.

-Phil

jazzed

Ken Gracey wrote: »

The infrared, magnetic tape and RFID systems are probably not as well suited for this application due to interference from people (for IR),

I wasn't suggesting using RFID as an RFID device

What I meant was that the reader could probably be hacked to make it into a metal detector.

I assume that the reader is programmable and that you have the source code (which ideally would be shared).

Genetix

I don't have an Arlo myself but my understanding is it is capable of a lot more than being herded in a ring.
Seeing an Arlo shepherd other Arlos, now that would be cool.
Having an Arlo speak when it sees a person or another Arlo, that would be cooler.
An Arlo with a personality, priceless!

garyg

bsnut
Which basic circuit are you referring to in post 18 of this thread. " a basic circuit is in the Forrest M. Mims, III book "Getting Started in Electronics" that sold in Radio Shack stores."
I have the 1984 printed version, so I may be missing out on which circuit you are referring to.
As far as I'm concerned at least, Ken Gracey is correct in saying there is interest in this type of boundary circuit.
I have a robot project that is on hold until I can figure out the boundary situation.

I've made several attempts that did not work out so good.
I learn something each time I make an attempt.
Low power radio or magnetic proximity detection gets very complicated to understand.
I will not give up.
My robot will stay within a wireless boundary.

Thanks for any comments.
gg

GordonMcComb

STEAM or STEM? Steampunk Arlos would be way cool!

The dog collar approach is probably the fastest and easiest. In the past I had some success with a wire follower using an AC wall adapter (AC output), and a couple 741 op amps. The dog collar probably has better overall range. The sensor loop of the homebrew follower had to be right over the wire in order to detect it, bit that was okay for the application.

Anyway, this sounds like a very nice project. Agreed this would make for a nice board-level product.

Phil Pilgrim (PhiPi)

GordonMcComb wrote:

STEAM or STEM?

Definitely STEAM: Science, Technology, Engineering, Art, and Math. My local school district has adopted the same acronym, in defiance of those soulless STEM advocates.

-Phil