Relative position in a field

Go To Last Post
23 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Over the Hol weekend I was talking through the night with a friend and the conversation drifted onto the strange lumps in his field. Inspired by watching Time Team, (or the excellent single malt) we reconned that we could 'easily' knock up something to do some geo-phys and map what's under the ground. His field is HUGE, but I can borrow RF equipment from work and we can tow it on a trailer behind his Land Rover, all good so far. I can easily knock up a little circuit to take the reading from the RF receiver and store it into FLASH until we download it into a laptop (rather than tow my laptop !). All good so far, BUT here's the problem, GPS only gives a resolution of 10m and we are probably looking to go better, say about 0.25m.

I am looking for any suggestions to map the location of the readings. I don't think 'dead reconning' is going to be any good as the Land Rover bounces over the field (it's pretty rough). If it was inside I would probably measure the position using ultrasonics, but there would have to be an ultrasonic PA stack to broadcast across the field, and think of the dogs in the village ! I don't want to broadcast RF as it's a pain getting it legal (other than the stuff we need to fire into the ground). Our fav idea (so far) is to put poles on 3 corners of the field flashing IR LEDs, then measuring the angles to them on the trailer, doing a bit of trig and working out the position.

We are looking for any easier ways of doing this. any suggestions greatly appreciated

Cheers :)

<º))))><

I am only one lab accident away from becoming a super villain.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

GPS accuracy amy be 10m globally but surely in one limited place on earth the relative displacement is much less? So if it's +7m wrong in one corner of the field it's surely +7m wrong in the opposite corner? But you might want to look at "differential GPS" which is something involving a local correction for position error I think?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

There is the spark fun comparison here: http://www.sparkfun.com/commerce/tutorial_info.php?tutorials_id=169
Most of these modules are spec'd under 3m....

The modules with WAAS, which is a form of differential, have better accuracy as well...
http://en.wikipedia.org/wiki/Wide_Area_Augmentation_System

It would be interesting to see what improvement having an additional stationary GPS as reference point could bring...

The GPS used for surveying seems to fit the bill, if you have the $$...

EDIT: My mistake, WAAS is only available in North America...
Your triangulation idea seems like it would work.
I may have to experiment with that for a grass cutting robot...

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Get a GPS made for walkers and rangers. These typically are made to register much finer displacements and slower speeds.

GPS by itself can be precise to a couple inches, and millitary implementations are. The limitations the civilian apparatus has are artificial. Governments don't want you to be able to strap a GPS on a rocket and send it in someone's chimney, like they do themselves... ;)

I am much more interested in the equipment you will be using to probe the terrain itself!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

try DGPS
A cheaper way would just be two GPS's one that don't move and one you do. If they are close and can see about the same part of the sky (use the same sat's) they will have about the same error, so you can calc. your position down to way less than 1 meter (if you can see more than about 5 sat's. )

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

sparrow2 wrote:
try DGPS
A cheaper way would just be two GPS's one that don't move and one you do. If they are close and can see about the same part of the sky (use the same sat's) they will have about the same error, so you can calc. your position down to way less than 1 meter (if you can see more than about 5 sat's. )

In this scenario, with a good quality GPS:

Autonomous GPS = 5m, $50 for bog-standard receiver
Autonomous GPS = 2m, $500 for good quality receiver
WAAS/EGNOS = 0.6m, $500
Differential GPS = 0.4-0.5m, 2x $500 receivers
Omnistar XP = 0.2m (requires subscription), $3K-$5K receiver plus signal subscription fees
L1 Carrier phase Differential DPS = 0.2m (Note 1), 2x $500 receivers, plus whatever the software upgrades are.
Omnistar HP = 0.1m (requires subscription), $3K-$5K receiver plus signal subscription fees
Real-time Kinematic (RTK) = 2-5cm, $30K total (base + rover + sw upgrades)

It goes up rapidly as the volume sold rapidly falls and the complexity of the software increases. And people who want accuracy are usually willing to pay for it.

Numbers are ballpark and will vary between manufactures and receivers. A good antenna is also a must, like a Pinwheel antenna instead of a patch antenna.

Once L2 and L5 signals become civilian and "open", I would expect the prices for hardware to fall. Current L2 requires funky cross-correlation of signals to recover the carrier phase of the otherwise encrypted code.

-- Damien

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Very nice breakdown!

When you refer to Differential GPS = 0.4-0.5m ...
Is that with dual Autonomous GPS = 2m or dual WAAS/EGNOS = 0.6m ?

SparkFun has modules for less than $100, some of which state accuracy at <2.5m
Their tracking comparison seems like a good read...
http://www.sparkfun.com/commerce/tutorial_info.php?tutorials_id=169
Any comments on the modules they stock?

Michael

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

krazatchu wrote:
Very nice breakdown!

When you refer to Differential GPS = 0.4-0.5m ...
Is that with dual Autonomous GPS = 2m or dual WAAS/EGNOS = 0.6m ?

Behind the scenes, WAAS **is** differential GPS, but by differential GPS, I'm referring to pseudorage corrections (i.e. measurement errors) from a stationary base being broadcast to a rover so that the rover can apply the corrections to its position solution. The stationary base (as long as it's close by) will supersede the corrections from WAAS.

WAAS is built for integrity ('can I trust the solution it gives me') rather than accuracy per se. The accuracy part is a nice by by-product.

Quote:

SparkFun has modules for less than $100, some of which state accuracy at <2.5m
Their tracking comparison seems like a good read...
http://www.sparkfun.com/commerce/tutorial_info.php?tutorials_id=169
Any comments on the modules they stock?

Rule 1. NEVER believe the marketing figures! They'll be stated for a particular set of conditions that may or may not apply in the area that you live in. Furthermore, some will optimise to use anything it can find in the sky and others will reject anything that looks fishy - it's a trade-off of accuracy vs availability.

Use one that has an external antenna. Try hunting around for a good antenna that rejects multipath, which I'd say is most of the garbage on the west-side of the building in that tutorial. At very least, get a patch antenna and put it on a large (say, 20-30cm) ground plane. On-board or integrated antennas are not that flash.

I'm loathe to make a recommendation without testing any of them, but Trimble has been doing survey-grade GPS for years and is likely to use at least some of that technology in its RF front-end and algorithms (and hence, the most stable result during that test). However, it appears not to support WAAS or differential corrections (look for "RTCM", "RTCA", or "CMR", which are common correction formats), so you may be out of luck. Maybe looks to see which ones support what features like that?

The alternative is to get the raw data (pseudorange, carrier phase, satellite navigation data) out of the receiver - the Trimble does it - and use GPSTK to generate what you want, but this solution does involve a lot of effort and you'll need to learn a few GPS fundamentals to do so.

-- Damien

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Great information, that certainly gives me a clear path in implementing a positional system...

My apologizes to Dren, the OP for hijacking his thread, I hope he is getting something useful from this discussion...

I was under the impression that the non-WAAS differential error correction was just a matter of subtracting error from a stationary known position...
That might have been an overly simplistic assumption as it seems there is more going on there...

I will have to look into this GPSTK ...
It seems like a differential positional system will be a project all of it's own...

I'm certainly not tied to any of the modules SparkFun offers, especially if most of them don't offer an external ant. connection ...
I will have to take a look at what other chips are offered by the likes of Trimble and others...

Thanks
Michael

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
Differential GPS = 0.4-0.5m, 2x $500 receivers

That is NOT what Differential GPS is! It's one receiver (GPS) that also listen to a local station (within about 1000Km)in the 200KHz band, and do the calc from that, and remember GPS is about 3 times more precise at nigth that in the day.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

sparrow2 wrote:
Quote:
Differential GPS = 0.4-0.5m, 2x $500 receivers

That is NOT what Differential GPS is! It's one receiver (GPS) that also listen to a local station (within about 1000Km)in the 200KHz band,

And what precisely do you think is measuring and broadcasting the corrections at the other end?

You can receive corrections via a number of formats. WAAS, as previously mentioned, is a special case of Differential GPS. Some maritime stations broadcast corrections from their own receivers, which is what I think is being referred to here.

Differential corrections may also be received via the Internet (NTRIP), often using a GSM/HSDPA connection out in the field. Sometimes these are generated from a single base, sometimes they are generated using a GPS network such as Leica SmartNet, used by surveyors and other precision users.

Sometimes a fee applies to receive corrections, other organisations will provide a stream for free.

-- Damien

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

krazatchu wrote:

I was under the impression that the non-WAAS differential error correction was just a matter of subtracting error from a stationary known position...
That might have been an overly simplistic assumption as it seems there is more going on there...

If and only if precisely the same set of satellites are received on both the rover and the base, and only if they are processed in the same way.

It only takes a single extra satellite to throw this out, though it should be better than an autonomous solution alone.

-- Damien

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

@krazatchu: This is fascinating stuff, I had sort of dismissed GPS, but I'll look back into using it and post up anything relevant I find. In the meantime I'll still carry on looking for any easier (cheaper) methods.

I don't want to throw money at this project, as it's just for fun. I mean we could hire professional geophys equipment if we wanted to seriously survey the field. To create a map I was thinking of measuring three things, the (electrical) resistance between the wheels, the refection of a ~1GHz Tx and possibly listening to the ground by either hitting it with a weight or blasting it with a low frequency audio signal. I'll post up this as a separate thread when (if) we eventually get to that stage.

<º))))><

I am only one lab accident away from becoming a super villain.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Dren wrote:
Over the Hol weekend I was talking through the night with a friend and the conversation drifted onto the strange lumps in his field.

How big is the field overall?

The IR idea is cool, but you'll have issues resolving height accurately, Maybe, just maybe, the horizontal resolution might be good enough, depending on how accurate the angle measurements are. At 100m, you'll need an angle resolution of 0.14deg or better to get 25cm over 100m. I think :)

-- Damien

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The field is roughly about 400m x 200m. TBH I haven't done any calculation on the angle resolution, and I was ignoring the height. I have a nice motor + encoder that could be geared to drive a turntable about 60RPM. It could could rotate an IR Rx and detect the peaks when it's pointing at one of the transmitters. I would probably flash them at xKHz and BPF the Rx circuit to block out ambient. LEDs are really good these days, so I feel that it would be possible to detect them over the distance even if we have to run a bank of them. I'll have to do some 'field' trials :) I suppose worse case we'll do the field in sections. GPS would be nicer if it can give the resolution for not too much cost.

<º))))><

I am only one lab accident away from becoming a super villain.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Dren wrote:
The field is roughly about 400m x 200m. TBH I haven't done any calculation on the angle resolution, and I was ignoring the height. I have a nice motor + encoder that could be geared to drive a turntable about 60RPM. It could could rotate an IR Rx and detect the peaks when it's pointing at one of the transmitters. I would probably flash them at xKHz and BPF the Rx circuit to block out ambient. LEDs are really good these days, so I feel that it would be possible to detect them over the distance even if we have to run a bank of them. I'll have to do some 'field' trials :) I suppose worse case we'll do the field in sections. GPS would be nicer if it can give the resolution for not too much cost.

I'm wondering about a laser rangefinder... Maybe something like this, but I don't know how well it works outdoors or at long ranges.

-- Damien

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

A quarter of an inch @ 30 meters gives three and a third inches at your longest dimension...

I suppose a few targets would be required and some method of tracking ...

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I think I originally misunderstood... I thought you wanted to scan what's UNDER the field and just correlate data with coordinates to map it.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

We do ? This is just the first step, to find out where the trailer is in the field so we can then plot the collected data in the right place. Ground resistance will just give us a 2D image, but the reflections (RF and/or Audio) should give an interesting 3D map of what's down there.
www.channel4.com/timeteam
Although TBH if we find anything at all we will undoubtedly find a series of low walls that archelogists will rant about then make up stuff about... In this room the Baron used to entertain his guests by juggling otters while an albino bear danced a waltz played on a lute made of solid gold and beeswax... oh really ?

<º))))><

I am only one lab accident away from becoming a super villain.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hey who knows maybe you got a spaceship under there...

On a more serious note usually surveying large area like this requires some segmentation, you don't usually do it one shot. Say you have an area 1KM by 1KM, you might wanna split it down in areas 10M x 10M and just do them square by square. Each area thus has absolute coordinates. This also makes it easier to track what has been done, and what's left to do.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

We are counting on it, and not a poncey 'scout' ship this time, we want the mothership or nothing !

Attachment(s): 

<º))))><

I am only one lab accident away from becoming a super villain.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

One cannot play a waltz on a lute made of solid gold and beeswax. DAMHIKT.

Four legs good, two legs bad, three legs stable.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Remember that GPS is very bad in the Z direction.

Instead of an encoder you could microstep of a step motor, that can give a very good resolution for cheap.
Raw numbers 200 step motor with 32 microstep give a resolution of about 0.05 deg.)
But whatever you do from the corners the calibration will be a challenge.