Source code for carrier phase DGPS using Allstar receivers..

[innovate1]

I will be putting together a more detailed introduction to the subject of Mowbots but, simply put, they are robotic mowers. Lot's of people have put them together with varying success and there is quite a bit of information on the net. The key is in the location determination and navigation software.

Several of us in the group went together and bought CMC Allstar GPS boards with the goal of using one as a reference position (stationary) and one on the mower. This should give us a position accuracy of about 6 - 8 inches - adequate for a mower.

The first site with detailed info and source code I found was:
www.precision-gps.org
This site gives some links to explanations of the math and such.

Another site of interest is:
http://mywebpages.comcast.net/dmilbert/softs/index.html
This one has source in Fortran but includes some info on the Allstar interface that the others don't seem to have. He didn't directly connect to the Allstar however. He logged the allstar data with a manufacturers program and his main program reads the resultant file.

I just relocated a site I ran across a few weeks ago. Looks like they did what we want to do with the GPS boards (not the mowing part).

Here is the homebrew Allstar GPS project info:

http://www.avrrc.lboro.ac.uk/essex/#label-19

Look under "Our GPS system"

I previously didn't find any source code or other details on their website and I dug around quite a bit. However, I just looked a little this time around and came right to it - links to the code, what it was based on, other notes - it's all there!

http://vase.essex.ac.uk/projects/gps/djjweb/

According to the page "The code has been tested on a Linux platform..." I think that makes you the expert, Lee... Want to check it out?

The other thing we need is antennas. Even if they aren't optimum, I would like to get something together to start testing. We can improve it as we go.

[LeenSTL]

It's a good sign that they don't get any compile errors, just some warnings. I'm not sure what to do with the code just yet tho ...

Here is a Gov't site which has a version of tar & gzip for Windows. At least for tar files (which are just archive directories & files, not compressed, usually have .tar suffix) this will extract the contents. Just do tar tvf somefile.tar to view the table of contents, if it looks good do tar xvf somefile.tar to extract the contents. It can even build a tar file that a Linux / UNIX system can read!

If a tar file has been gzipped (usually has .gz as a suffix) first then use gzip to ungzip it then use tar to extract the contents.

[innovate1]

I added a few of the primary links for program info to my sticky message but notice it doesn't show up as a new message so I am posting this note about it. Since I don't seem to find the time to get all the info together at one time I will be doing incremental additions of info as I have time. Others should feel free to add as well.

Phil, Can I move some messages about mowbots from another section to here? I don't think I have enough permissions so I might need to ask you to do it.

Jim

[phild]

let me know which topics and I'll move them

[mhamon]

I've been thhinking of a GPS based guiding system for my mowbot; and so far the expected precision seems to be not enough except with the carrier phase DGPS ! but it's seems to complicated for me; has anyone of you go farther in this way ?
Have you ever tried to put all the necessary code in a microcontroller ?

[Carlton]

I'm currently working on a DGPS system at my house (as a back burner project)
from the litature prior DGPS systems can't get better than +/- 2ft, if you have a GPS lock with both WAS satalites you could get +/- 5 ft on WAS

so depending on how much edge you can leave and go weed-eater afterward, both could be good enough or both could be unacceptable as the mower itself is less than 2ft wide

the other idea I've played around with is forget GPS and use 2 atomic clocks with 2 transmitters in fixed location and have the mowbot with an atomic clock do its own time time/phase shift analysis of its position. if that was coupled with an electronic compus and well calibrated,
I starting to think this 3rd option may be more accurate for my small yard.

[mhamon]

It's sound good but quite out of my level ! I've found on internet another way to compute the mowbot position:
it's base on several base which are equipped with both IR transceicer and US transmitter; the mowbot is also equipped and always scanning the the available IR base return.When an IR communication is established between the mowbot and the base; then the base sends a US pulse, the time to travel is then quite easily to measure, thus the distance to the position to the base.
has any one tried this ?

[Carlton]

wether you use radio wave(as I suggested eariler) or infred light(which could have interference outdoors),
both mve at the speed of light = 300,000,000m/sec or ~186,000mile/sec in a vacuum

I just realized that both are subject to the sampling rate of your controller or timing device

Clock Speed __ distance resolution (meters)
at 1Mhz _____ 300m
at 5MHz _____ 60m
at 8MHz _____ 37.5m
at 16MHz ____ 18.75 m
at 50MHz _____ 6 m
at 455 MHz ___ .66m

I don't know if you could loop the ping as a delay to divide the distance for better position distance

[mhamon]

that why this system uses both IR and US (UltraSound) that moves at a much slower speed say 330 m/S thus the mowbot can compute the distance by starting it's counter at the reception of the IR pulse and stopping the counter when receiving the US pulse

[Carlton]

mhamon,

Do you live in France? you don't have a location posted in your profile

yes, your right if UltraSonic can work it would be much better in resolution.

I am trying to think out how to add self-stearing to a gasoline self propelled lawn mower, so don't know if the microphone could hear over the engine noise

[mhamon]

Yes I live in France, close to Nantes quite far from you ! but internet doesn't mind Km , sorry Sm

[mhamon]

At least I found a very good study of an IR/US beacon at http://www.convict.lu/htm/rob/ir_us.htm.

[innovate1]

mhamon,

We never got too far with this. The math and programming are pretty intense. The accuracy of 20 - 50 cm is marginally good enough for a mower I think. Since this requires a radio link from the stationary GPS to the moving GPS I wasn't worried about putting this into a micro. I figured the main number crunching could take place on a PC.

The IR-US method has some interesting possibilities.

The aircraft VOR system in this country (very old but still active) uses an RF system to yield angle from the transmitter. Several stations are used to triangulate position. Would probably work ok in small areas but accuracy would suffer in large ones. Might be possible to do something similar with IR.

Jim

[mhamon]

A very precise navigation system would enable a "parallel mowing" and the boundary detection would be simplified.
The very precise gps based on phase calculation may be precise up to 1 or 2 cm! no other system would be necessary.

As a pilot I well know the VOR, DME, ADF and now GPS based navigation system but these systems are only radio VHF and UHF.
Ultra sonic moves much slower and computation are then made easier for distance lower than 10m.

That why I think there only 2 ways : 1)High accuracy GPS 2)ground based system using US and IR or radio

so let's try !

[Carlton]

Jim that give me a great Idea, Thanks for bringing up VOR

The mower could have a transmitter on it.
Let 2 or 3 well spaced recievers (at percisely seperated distances) calculate the angle the mower is from north (I'd probably use magnetic), then transmit that angle data back to the mower reciever each in turn.

the mower controller would then using law of Cosines, knowledge of where it was, and an electronic compass it could exactly find where it is and when it needs to turn next and which direction

This would reduce the sensor equipment on the mower to a tranciever and a compass.