Magnetic wire to define border area for robot path

[marcusbarnet]

Hi to all,

I have a mobile robot which is able to move randomly outdoor on sandy and dusty grounds and I need to find a solution to correctly define the area within it has to move: in particular, I have to mark the area borders so the robot can't come out the defined area.

I already tried sonars and IR solutions, but they are not suitable for my application because i'm outdoor and the ground is sandy and dusty; moreover, my robot moves under the sun so IR sensors can't work very well.

I'd like to use a magnetic wire (or RF) (may be to lay it underground) to define the area borders and let my robot be able to detect the magnetic field when it comes near the border in order to implement something like this:

- when no magnetic field is detected, robot moves randomly in the area
- if magnetic field is detected, so robot is near the border and it has to change its path

The wire has to be 60-80 meters long, at least, because my robot is big and so the area to cover.

Can you give me some suggestions on how to realize the magnetic or RF wire, please?
I can't figure out the circuit to use (I think i should have to use a NE555).

To detect the magnetic field, I'm thinking to use the MicroMag 3-Axis sensor (or something like it).
What kind of other components I need to develop my solution?
Can you help me, please?

(I'm using Arduino Mega to control my robot)

 

[3v0]

Look into the electronic dog fence technology. I would expect that you can find some DIY info on them on the web. I recall one guy using it for a robot mower.

 

[marcusbarnet]

Thanks a lot for the suggestions!

I searched on google for this kind of documents, but i've found nothing

 

[3v0]

This guy figured out how to do it.
https://www.youtube.com/watch?v=zuXqg4q3cQ8

 

[marcusbarnet]

Yes, I'm thinking to buy an electronic dog fence and to use the transmitter as it is and then to modifying the receiver to trap the signal during the detection.
I hope this can be done: may be the manufacter has protected the receiver and it can't be disassembled.

 

[3v0]

The system uses a radio signal.

According to what I read the unit worn by the dogs beep then shock. You could tie into either of these without opening the case if needed.

 

[marcusbarnet]

Yep, I hope so!

I'll buy one later and I'll let you know if it'll work!

 

[panic mode]

in couple of plants they use underground (in the floor) wire to let the autonomous transports navigate the plant.
the 'wire' is actually 7-10 conductor cable, wires at cable ends are connected to form coil which is attached to audio amp.
amp is driven by small oscillator (4-5kHz I think). cable was buried about 8" and slot (cut by double blade diamond saws) was filled by epoxy.
transporters have two pickup coils mounted low and connected to circuit board that controls steering.
there ware other guidance options too so the transport could be sent (purposely) off path, like detour etc.
it would sense surroundings, avoid obstacles and eventually converge on path again.

 

[marcusbarnet]

Thanks a lot!

I think the RF cable is the correct solution for my application..

 

[cr0sh]

I've thought about something like this, but I didn't want to lay a wire; I have a Friendly Robotics RL-500 mower that someday I want to get working again, but I want to replace the controller and other parts of it, and not use the original wire border sense interface. My "thought solution" is to use small rare-earth magnets epoxied or otherwise affixed to some kind of small plastic spike, inserted into the lawn at regular intervals around the border. Then use hall-effect sensors on the robot to detect the border. If the idea worked, it would be cheaper and easier to implement, plus no need for power to a buried wire. Also, unlike a wire, it won't break (it may be difficult to find that break later).

 

[marcusbarnet]

Yep, it's a great idea!

But the problem is where to find these rare-earth magnets epoxied: I 've never seen them neither online nor by my local dealers.

Can you give me any input on where to find them, please?
How much do they cost?

 

[3v0]

cr0sh how close together are you thinking the magnets need to be.

 

[marcusbarnet]

The problem might be the presence of metal pieces near the magnets...

 

[4pyros]

Yep, it's a great idea!

But the problem is where to find these rare-earth magnets epoxied: I 've never seen them neither online nor by my local dealers.

Can you give me any input on where to find them, please?
How much do they cost?

Old disk drives.

 

[marcusbarnet]

But, it is not possible to use old magnets from hard disk drives if you want to cover an area of 100^2 meters or more..

 

[cr0sh]

Just to answer some questions:

I would buy the magnets off of ebay; they are very cheap (just search "rare-earth magnets"). For the spikes I would probably use plastic golf tees. Just epoxy (or silicone?) the magnet into the divot.

For my lawn, I was thinking 6-8 inch spacing; granted this is a small residential lot - it wouldn't take long to place these around the perimeter. I doubt that metal near these passive magnet spikes would have a great effect, given their strength, though if your robot had a metal structure it might be an issue (my mower robot is mostly plastic).

Covering a large perimeter would be time consuming, but not impossible; you might hire some people to help place them, you might make a machine or other device to help place them. You might just suck it up and get a sore back placing them all weekend (or more) long.

My idea, though, was for a small residential setting; I was going to place a set of hall effect sensors around the front edge of the robot to sense the magnets. It already has a bump sensor of some sort (that, or it used current monitoring of the motors to detect resistance in it's standard configuration - the RL-500 worked something like a Roomba, using a random-walk style system with a buried perimeter wire). My current stumbling block in my thoughts of how to refurbish it is I need a sensor to determine cut grass from uncut grass (as well as possibly non-grass). So far, I've been looking for a high-speed humidity sensor (cut grass has a higher humidity than non-cut grass; I wouldn't be using this mower on a freshly watered lawn, either). Another more complex possibility has been the idea of a vision system (waaay more complex - but possible). I've had a few other ideas as well, but the humidity sensor seems like the best option (except so far "high-speed" hasn't translated into "under 1 second response time" in my searching).

Ultimately, what I would love this robot to do is cut the grass with as little of "random walking" as possible, sense cut areas from non-cut/non-grass areas (and not turn the cutter blades on for those), and generally be intelligent about where it has been and where it has to go (ultimately getting a fully cut lawn, while reducing the usage of the battery).

 

[4pyros]

But, it is not possible to use old magnets from hard disk drives if you want to cover an area of 100^2 meters or more..

Why not?

 

[4pyros]

Ultimately, what I would love this robot to do is cut the grass with as little of "random walking" as possible, sense cut areas from non-cut/non-grass areas (and not turn the cutter blades on for those), and generally be intelligent about where it has been and where it has to go (ultimately getting a fully cut lawn, while reducing the usage of the battery).

Maybe GPS?

 

[cr0sh]

Why not?

Based on my limited experience with opening hard drives (admittedly, nothing recent, but I doubt things have changed much), there are two styles of magnets (discounting those in the motor, of course) in hard drives:

One is a semi-circular "D" shaped magnet below the head positioning actuator coil; depending on the age of the drive, there may be one or two present (if two, one will be above, the other below), each mounted on a metal bracket with some kind of epoxy/super-glue or something, and the brackets are spaced from each other (and the coil) by either metal/plastic separators or molded in tabs on the metal brackets (part of the stamping process, I imagine); sometimes it is a combination. You'll usually find two magnets on older drives from the 1990s or so; I believe they used alnico or some other magnet technology, and thus needed to double up on them to make the magnetic gradient stronger (?). Later drives used rare-earth, which are much stronger (but more fragile). The magnets are essentially bar magnets with a north and south "end"; the coil changes it's magnetic field orientation amount to move the actuator along this gradient.

These magnets are fairly large, but that might be ok on a large perimeter; the only issue would be how to mount them to keep them from being hit by blades or dislodged in some other manner. Probably the best way would be to use a small trencher and bury them down 6-8 inches under the surface. There's also the issue of removing them from the metal brackets; you can use a hammer and screwdriver (or cold chisel) to do this, but you'll break as many as you release, generally. For my needs, this was too much effort for too little gain (might as well bury a wire). Which is why I'd rather purchase my magnets and golf-tees for the purpose.

The other type of magnet, which isn't in all hard drives, btw, is a small little "keeper" magnet (the thing is tiny, maybe 1-2mm in diameter?, and nearly impossible to remove from the plastic piece it is embedded in). It's purpose (I think?) is to, as part of some kind of electromagnetic assembly (?), hold the drive's actuator arm away from the platters on power-down. The arm on power-down quickly swings over to it's "parked" location, and this device latches it in some manner, so that the arm can't move when it is transported or otherwise handled. Generally they are not worth keeping when disassembling a drive, as removing them from the plastic bit can either be fairly easy, or nearly impossible, depending on the drive.

I learned all of this while, for one of my previous employers, disassembling a bunch of old drives for "drive destruction" (which was part of his contract with his clients); I gave him all of the metal platters and spacer rings (he wanted to make some wind chimes or something), I kept the motors, magnets, drive cases, and miscellaneous screws. I have a whole box of this junk. The drive cases, one day I intend to take to be recycled for the aluminum (and I should do it soon, as who knows when or if the bottom will fall on prices for the metal; though for now all it does is seem to rise). I won't get rich, but I might get enough money for a tank of gas or something.

 

[cr0sh]

Maybe GPS?

Ultimately not accurate enough; DGPS might be, but I don't have that kind of money. Plus, I intend to use this mower mainly in my back yard, which has a huge (40 foot tall) mesquite tree that covers the entire yard; GPS can't see the sky with it in the way. So, GPS is virtually useless.

Fortunately, though, I am currently taking the Udacity "CS73: Programming A Robotic Car" being taught by Sebastian Thrun (**broken link removed**), which aims to give you the skills to "Learn how to program all the major systems of a robotic car from the leader of Google and Stanford's autonomous driving teams".

These skills should translate well for controlling a robotic mower in a more intelligent manner; I figure between this course, and what I learned in the free Stanford ML Course last year, learning how to localize a robot within an environment should be fairly doable. The backyard is fairly static, so once I have the map, and some basic odometry measurements, coupled with some measurements from some sort of distance sensor, there's a good chance it could all work out.

Hopefullly.