SparkFun Forums 

Where electronics enthusiasts find answers.

Tips and questions relating to the GPS modules from SFE
By Oznog
#44112
Realistically, if I have two GPS modules- one at a known point broadcasting a differential correction to a roving module and take readings around the yard with the roving module like once every 2 sec or so, what kind of accuracy can be expected? I assume some of these modules would be much more accurate than others for this job?

Trees may be an issue, it may also be near a house which causes a bit of reflected signal.
By NleahciM
#44123
Oznog wrote:Realistically, if I have two GPS modules- one at a known point broadcasting a differential correction to a roving module and take readings around the yard with the roving module like once every 2 sec or so, what kind of accuracy can be expected? I assume some of these modules would be much more accurate than others for this job?

Trees may be an issue, it may also be near a house which causes a bit of reflected signal.
I'd advise taking a look at this website: http://www.precision-gps.org/

It should get you headed on the right track.
By jmdesign
#44124
I think realistically, you're going to still see the base 3m typical resolution. You might be able to get it a little bit closer using a Kalman filter and combining what you've got with the known position, but I don't think you're going to see any major improvement. You may want to combine the GPS with another distance triangulation method (sonar, radio beacons, etc.) which is more reliable in your environment.

I'm working on something similar for accuracy improvement.
By NleahciM
#44160
jmdesign wrote:
NleahciM wrote:I'd advise taking a look at this website: http://www.precision-gps.org/

It should get you headed on the right track.
This is interesting, I haven't seen this before. I'm going to take a look at their approach...
It's nothing out of the ordinary. The main mistake alot of people make when trying to do differential GPS is to just subtract how much a non-mobile GPS device is moving from the mobile GPS's position. This is fundamentally a terrible thing to do, as it would be incredibly silly to assume that the two GPS devices are using the exact same satellites. (among many other silly reasons)

The math to do it the right way is fairly complicated - but luckily it's nothing new... There are plenty of resources on the topic out there.
By Oznog
#44310
NleahciM wrote: The main mistake alot of people make when trying to do differential GPS is to just subtract how much a non-mobile GPS device is moving from the mobile GPS's position. This is fundamentally a terrible thing to do, as it would be incredibly silly to assume that the two GPS devices are using the exact same satellites. (among many other silly reasons)
Yeah that link does have some interesting info for sure.
I'm still not sure what kind of error I could expect with a nearby (<300 ft) reference station though. Or are the differential corrections broadcast from the towers actually comparable in the kind of accuracy boost they give?

I should probably stop being cryptic, after all I get on other people's cases for that when they ask for help without saying what they need it for. Basically I was thinking about converting a Friendly Robotics Robomower, a self-propelled and guided electric mower, which uses a buried wire, like the Invisible Pet Fence, for boundaries and kind of bumps back and forth randomly off obstacles and perimeter. Uses a magnetic compass to stay on heading once it starts in a line, but the actual direction is not chosen with great care. The process is random and to actually ensure the entire lawn is mowed without a single unmowed spot takes enough runtime to do the lawn many times over.

It would be an interesting project.
By ttabbal
#44378
If you keep the existing wire and compass guidance you should be able to add GPS even with the 3m resolution and get smart guessing out of it. It should cut down on mow time even if it's not 100% accurate. You can know that you probably need to turn to a 200deg heading and use the compass to guide you. Figure on hitting everything twice to help with GPS inaccuracy and you should get pretty good results.
By Oznog
#45752
Well, I'm looking into this differential project more. I'm probably going to have the MTK MT3318 GPS module. I see that every satellite broadcasts low accuracy "almanac data" on every satellite's path and more accurate "ephemeris data" that only pertains to the path of the satellite that broadcast it.

So I can make this module output this data for external calculation purposes? What data am I looking for then that gives the actual distance/cycles/phase of the positioning signals received from the satellites? I've been looking through lots of difference pieces of documentation to feel this project out but I don't see where this actual carrier timing data is made externally available, which would make the project pretty much impossible.