need advice on sensing buried cable robot

hax

New Member
So we are building a house that needs a septic system installed, which involves sub-surface irrigation pipes, and a grass field that must be mowed regularly. (17metres x 18metres)

I want to build a lawn mower robot ;-) , to mow the lawn automatically. At this stage we are laying the sub surface irrigation pipes and I think it's a great opportunity to lay a cable in with the pipes. I have found some "direct burial" Cat 5 cable that I am thinking will work well if I connect all of the conductors together at the ends. It's gel filled, so it should last some time.

The question to the collective group here, is, how do I sense the cable?

I have an idea, but if you think I am waaay off, let me know.

I have an RF sniffer (cable finder) that I purchased from ebay which works well. It has a transmitter that connects to a cable, and a negative crocodile clip which you clip to earth. The receiver wand seems to just be a coil with a very high gain audio amplifier. If I wave it over a power line, I can hear the 50Hz hum. If I wave it over the RF transmitter, I hear a dual tone, and interestingly, if I wave it over an analog telephone line, I can hear the coversation!!! :0)

So I am thinking of sending a single DTMF tone over the "transmitter" and using a series of pickups in the robot, with DTMF receivers, to listen to the power line.

What do you think? Any better ways? I have not tried it under ground yet. I wonder if the ground itself will act like an earth and shield the signal, especially since its around 100m of cable.

It's got to be reliable of course, as I don't want it to drive away, hence I think the DTMF tone adds some reliability.

I could simply connect the two ends of the cable to an audio amplifier, that way it would make a large electromagnet. Probably easier to pick up than RF?
 

SAborn

Senior Member
Interesting project, i dont know the answer to your question but will follow the thread out of interest.

The lesser non electronic solution would be ..get a goat, but the picaxe would be harder to retrofit then.
 

westaust55

Moderator
There are some very good professional buried cable and pipe locator units such as the CAT series by Cable Detection. See: http://www.cabledetection.co.uk/ezicat-i550

These are in self contained mode (without a separate RF generator) are quite accurate and we have used earlier models with the “Wand” held vertical for the location and then held at 45 degree to horizontal to determine the depth.

I guess that the concept is not greatly different to metal detectors used by gold prospectors and tourists wandering through the countryside around old gold mining towns (at least here in Australia).
Possibly using two home constructed buried cable locator systems as either side of the robot mower and angled inwards “V” fashion could be used as a form of line follower.

Maybe add some form of signal strength to determine of straying (and GPS to find its way home if it locks onto a telephone cable running down the street?)
 

SAborn

Senior Member
if it locks onto a telephone cable running down the street?)
That could be rather funny. :D leaving nicely mowed strips through Mrs,Jones prize flower garden down the street, and any other neighbours garden.
 

Jeremy Harris

Senior Member
I did this on a smaller scale years ago, although I never managed to get the robot to cut the grass well. The sensing system I used worked well and didn't need the cable to be buried (although it could have been, very easily). At the time I'd been working on low frequency cave communications (low frequency RF goes through soil and rock to some extent).

The ground cable was a single core loop, fed with an AM modulated 100kHz signal. I used the mains frequency to modulate the 100kHz transmitter, which was just a simple crystal oscillator with a FET power stage. By applying AC to the power stage, overlaid on the power supply with a small transformer, I was able to get around 80% modulation depth.

The receiver on the mower was an old ferrite aerial, taken from a portable radio. I used the LW coil, tuned to 100kHz, and one of the cheap three pin ZN414 tuned radio frequency receiver chips. The output from this was fed to a comparator and then I just used a phase locked loop frequency detector to detect the modulation frequency.

In practice the receiver reliably detected the wire from around 200 to 300mm away, so I just arranged for it to be staked around the edge of flower beds etc.

The mower was the weak point. I got the vehicle to drive around randomly easily enough, using a simple control system (using another type of ucontroller) that used the RF sensor and two collision sensors (microswitches mounted to arms at the front either side). The algorithm was simple, just drive forward until a sensor was triggered, then reverse for a set time, turn through a set angle and move forward again. In practice this gave a pretty random movement that covered the whole lawn after a time.

The major failing was the cutter. I opted for a technique whereby the mower would run all the time. It had a solar panel for power, charging a sealed lead acid battery. The cutter was a rotating blade with a Stanley knife blade set in the end and was intended to just cut a few stray blades of grass at a time (it was designed to literally trim the grass as it grew, rather than wait until after it had grown a fair bit). I never managed to get enough traction from the wheels when the grass was wet and also never resolved the lack of power available at the cutter head.

I concluded that I needed a much bigger, more powerful unit, with a normal cutting strategy. After a whole summer of playing with it I gave up!
 

lewisg

Senior Member
The cable locator I use (the old Ideal version of this) nulls over the cable. It works like the unit described by Westaust55. Find the cable horizontally then holding the probe at 45 degrees find the null again and that is your depth.

Using a device like this to guide a robot mower could be done like a line following robot.

Other methods would be to surround the area to be mowed with a signal cord and have it "wander in the enclosed area. Or it could follow the perimeter then once it got back to the origin "step" in the mower width and make another loop.

If you don't have much wheel slippage you could do the whole thing with counting wheel revolutions against a known path. At that point your perimeter wire is just a safety feature.

I'd look at using the wire they use for those "wireless fences" for dogs.
 

John West

Senior Member
I'd try running a low frequency signal through a back and forth wire layout, then using the sort of loop detection system mentioned above to determine signal strength (with a notch filter to limit reception to just the transmitted signal,) the mower could follow the maximum signal. I'd add a 'minimum signal strength' cut off that would stop the mower if the signal dropped off too far. With that in place the mower would be reasonably safe from wandering away.

I've worked on computer-controlled army tanks, so I do understand the concern for making sure the device doesn't go berserk and wander off. In fact, I'd suggest you consider building some sort of fence around the area you wish mowed that could stop a run-away mower. It might not take much in that regard, but any control methodology implemented should be made as fail-safe as possible with regards to limiting uncontrolled wandering, and have back-ups to guarantee it.
 
Last edited:

geoff07

Senior Member
My point also. Pet collar shocking systems are illegal now in the UK but you can still get the bits in some shops, and they may still be legal wherever you are.
 

Paix

Senior Member
I didn't know that about Pet collar shocking systems. Next it will be stock electric fences being outlawed . . . where will we be then, or more to the point the horses and things that some rely upon them to control?
 

boriz

Senior Member
Idea:

Cheapest/simplest solution I can think of. Prolly be the one I'd try first:

Two 08m Picaxes. One is the TX, one the RX.

TX: Drive the fence loop with an 08m + MOSFET using the IRout commands. This provides a nice coded 38Khz signal.

RX: Use a coil and transistor amplifier (probably will not need to be tuned), with the signal driving an IR LED which is directly coupled to an IR RX module (like this one: http://www.techsupplies.co.uk/epages/Store.sf/en_GB/?ObjectPath=/Shops/Store.TechSupplies/Products/LED020). The IR module has a nice demodulator in it, and drives the RX Picaxe directly, making the rest of the hardware simple.

The TX and RX can be set for a unique code that will eliminate false triggers etc and supports the possibility of using different codes for different boundary lines, making navigation and mowing patterns easier to control.

This is not an RF system. It's an induction system that should give good communication over short range while not effecting or being effected by external noise.
 

fritz42_male

Senior Member
I'd go the organic solution (the goat)

Low tech, high reliability, long product life under normal conditions (8-10 years), solar powered (well ultimately it eats grass which needs light to grow) and the waste products are beneficial for the lawn.

Nigerian dwarf goats.
 

SAborn

Senior Member
Then perhaps the goat with a shock collar, involves the best of both designs, Nature and the picaxe.
It can also be saftey teathered, no traction problems, and the program logic is very simple, (water,food), you could call the goat Victor. (common brand of mower here)
 

techElder

Well-known member
Solving what problem? That yard isn't even 400 sq. meters. or less than 1/10 of an acre. Dang we mow 3 to 4 acres every other week! :rolleyes:
 

fritz42_male

Senior Member
Then perhaps the goat with a shock collar, involves the best of both designs, Nature and the picaxe.
It can also be saftey teathered, no traction problems, and the program logic is very simple, (water,food), you could call the goat Victor. (common brand of mower here)
lol @ Victor

Yes the program logic is simple until you hit an error condition and then it's a bugger! That's the trouble with organic systems.
 

hax

New Member
Thanks guys, lots of good info here. I'm tossing up the options now of an rf vs induction system. Not sure if the rf system will work well through the ground as after all, the ground itself is also the rf ground too. I guess it's time for a real world test. Got to get the hands dirty. I'll start with a long cable and connect both ends to a standard audio amplifier and see if my cable sniffer is able to pick up the audio signal. I'll have an 8 ohm resistor in series if the resistance of the cable is too low. Will let you know how it goes!
 

boriz

Senior Member
Use a multi-core cable and solder each wire to it's neighbor at the return. IE:

Imagine a two core cable of 1 meter length. If you name the wire ends as A and B at one end, C and D at the other, then if you connect C to B you have a single conductor of two meters.

If you extrapolate this, you can use a multicolor cable to get a very long single conductor in a big loop (coil). The total resistance may be high enough to drive directly with 1A pulses.
 

Paix

Senior Member
An inherrant problem with the Victor mower is that as well as being very efficient at mowing the lawn, there is occasionally the phenomenon that it's smart enough to be able to service your car and daft enough to service the OH too. Click . . . BANG is increasingly frowned upon as a remedial course of action in such circumstances :)

In lonely outlaying areas I would also steer clear of the young Billy model for similar reasons, although the chances of of a similar malfunction are less remote and the Click . . . BANG solution that much more acceptable, particularly if you are fond of curry.

A Picaxe solution does sound a lot more attractive than some of the off the shelf systems mentioned, now that I come to think about it.
 

SAborn

Senior Member
Its good to see that some members still have a sense of humour with a little diversion of OT, elseif you love curry or not.
Sorry if i offended anyone with ...Click.....Bang, but i did grow up on a farm where reality was often harsh.
 

hax

New Member
Boriz I don't think your idea will work as the magnetic field will cancel out. You will have two currents one way and two identical currents the other way. But I'll try a few things and get back to the group.
 

Jeremy Harris

Senior Member
Thanks guys, lots of good info here. I'm tossing up the options now of an rf vs induction system. Not sure if the rf system will work well through the ground as after all, the ground itself is also the rf ground too. I guess it's time for a real world test. Got to get the hands dirty. I'll start with a long cable and connect both ends to a standard audio amplifier and see if my cable sniffer is able to pick up the audio signal. I'll have an 8 ohm resistor in series if the resistance of the cable is too low. Will let you know how it goes!
LF RF works great through the ground. I've used cave comms gear (working at 80kHz) from the surface to underground through several tens of metres of soil and solid rock. Making it work through a few cm of soil is trivial and only needs a tiny amount of power. The ferret locator I built years ago used a 100kHz transmitter collar on the ferret, with a tiny ferrite aerial, and a loop wire directional receiver aerial on the surface. That also worked well at a depth of several feet under the soil (we used it to find where ferrets were in rabbit warrens when they stayed down, so we knew where to start digging to retrieve them).
 

hgestel

New Member
Does anybody have more information hardware and software of the Robot Lawn Mower project of Tobias Simonsen.
It would be very nice if someone is prepared to share this info as I can not reach Tobias in any way.

Thanks
Henry
 

sid

Senior Member
There are some very good professional buried cable and pipe locator units such as the CAT series by Cable Detection. See: http://www.cabledetection.co.uk/ezicat-i550

These are in self contained mode (without a separate RF generator) are quite accurate and we have used earlier models with the “Wand” held vertical for the location and then held at 45 degree to horizontal to determine the depth.

I guess that the concept is not greatly different to metal detectors used by gold prospectors and tourists wandering through the countryside around old gold mining towns (at least here in Australia).
Possibly using two home constructed buried cable locator systems as either side of the robot mower and angled inwards “V” fashion could be used as a form of line follower.

Maybe add some form of signal strength to determine of straying (and GPS to find its way home if it locks onto a telephone cable running down the street?)
I'm phone engineer in the real world and we use the cat and genny to locate buried phone cables in the ways described here and before that we used a box of tricks that put a powerful tone down the cable and then used an induction coil to locate said cable. both work but both have limitations and both will pick up on any other metal work in the ground, eg power cables, water pipes etc and the type of ground also affects how well these devices work. However all that aside I think westaust55 found the answer when he mentioned gps. large agribusiness now use driver-less combine harvesters that drive up and down 24hrs a day and night all controlled by gps. perhaps you could use a picaxe as an interface between a gps and the mower, in fact I expect if you have deep enough pockets there are already mowers or units available on the market that do exactly what you want. If you go down the gps route at least you'll have accuracy.
Rgds,
Sid
 

papaof2

Senior Member
If you go down the gps route at least you'll have accuracy.
That depends on whether you can find a better-than-1-meter-accuracy GPS receiver and whether you can afford it if you do find it. Garden variety GPS units are typically 10 meter accuracy and aren't likely to provide guidance for straight line grass cutting in typical mower cut widths.

Try this: select the GPS page that displays the lat/long values and see whether
1) it shows a constant value when not moving
2) shows a change of 1 meter when you move it 1 meter in any direction and let it stabilize there.
 

Paix

Senior Member
@Papaof2, I think you need a lawn mower from Klaas . . . eek, but watch the pansies.

I believe that the accuracy can be improved immensely through the use of a reference receiver, so that a known fixed location GPS receiver signal is compared with that of the mobile GPS and accuracies down to 10cm can be achieved, but beyond understanding that, the actual means of achieving it is far beyond my capabilities.

Best if the Klaas machine isn't connected to the RBS computing systems though, given this past week's experiences, if you value the flower bed :)
 
Top