I've read through a few discussions on this forum about poor man's DGPS and getting relative accuracy down to the ~1m range. It seems no one's having much luck even with identical GPS units. I'm wondering if anyone has manged this and what are your thoughts on what I'm currently building.
I'm laying out a PCB for a ublox LEA-6T (that's the one that outputs raw data and carrier phase) to interface to a MCU and a 433MHz radio. That's the mobile unit. My base station is then a single board computer with it's own LEA-6T, radio, and software running GPSTk post-processing software. The mobile board will then radio the raw data (subframes 1 and 2 and psuedo range and carrier info) to my base station. Base station code then chooses a set of sattellites that work well for both units and calculates position of both units using the same satallites and same ephemeris data. (calculated position could then be radioed back to mobile unit)
The units are genearlly pretty close to each other (spatially) so the only error here should be that inherent to the ublox measuring psuedorange--no constellation, algorithm, iono/troposphere, ephemeris differences.
Has anybody tried this? Do you think I can get a relative accuracy of ~1m?
Thanks for any thoughts guys and gals!
Interesting... would you be forming distance measurements with theodolite or just working from angular measure off the optic/photo's? You may have to carefully calibrate the optics of you system to account for distortion/etc.
Once RTK (or post processing) is 'fixed' the instaneous readings should give you a good position. If you are taking 'snapshot' pictures you may want to use a 'mark' input or record waypoints.
I believe that the only choices in 'consumer grade' are Ublox5/6-T, SkyTraq (Raw) and DIY.
Stepping up to commerical look at Novatel, Topcon, Leica, etc... though lots of money. The quality of you antennas will probably be the largest factor in accuracy.
probably the best place to start for SfM is either visiting the photosynth site or downloading the free StereoScan by AgiSoft. This ceates a 3D point-cloud automatically from just two photos. All distortions etc are taken care of and greater accuracy can be achieved by making a camera calibration. The GCPs allow you to scale and orientate the final product to the desired grid and depending on what additional processing software you have available it will make any further adjustments to produce accurate orthophotos. PhotoScan Pro is such a product. All I need now is an accurate method of surveying the GCPs - low cost RTK such as we are discussing is probably sufficient.
Will try the LEA-6T - does it output serial? Other than the U-Blox evaluation kit is there a source of these mounted on a PCB either with an antenna or somewhere to connect one?
The eval kit is probably your best 'get going quick' solution, a bit pricy at $350 and you'd probably want to get a better antenna (the patch is a bit crappy). At least you don't have to fuss with a 'carrier' PCB and level shifting/etc.
It does support RS232 serial (DE9 hidden from view on other end of enclosure) and can be power from an 'ipod' USB supply (wall wart or cigarette lighter output).
Skytraq kits are available too, but don't come in a enclosure.
I have several Lea-4T mounted on professional multilayer circuit boards about 30 mm x 65 mm, complete with voltage regulator plus antenna short circuit detection, and smb antenna socket. They have serial output (logic levels) and fairly easily wired up for direct off usb operation, including power from usb if required
This is basically the guts of an evaluation kit but bareboard.
Also has GPS timing output.
Also have details of cheap rtk able antenna system I developed. Its basically helical antenna plus home made choke ring. I am getting better results than say trimble 16741 or Novatel 501pm me if you are interested.
Be warned before you start a design with LEA-6 that I have had a hell of time getting reasonably priced low quantities in North America. ublox America sells the non-raw output models (like LEA-6S and 6A) for $100 each. The only international distributer I could find that would ship to Canada was http://www.rfdesign.co.za/ and they charge $30 for the S and A and ~$100 for the T but $100 shipping to Canada. Ublox seems to keep tight control on their distribution and really adjusts the prices based on region. There are some break-out boards with LEA-5 on them but I haven't found any with 6.
If anyone knows of a cheaper source for low quantity ublox, please let me know!
I just designed a board for myself and you can have the Gerbers if you want:
-LEA-6(S, A, or T) with antenna connector (I-Pex MHFI)
-filter caps and ferrite chips to quite the input to LEA
-PIC18F2580 (or 2480)
-ISM radio from HopeRF (http://www.hoperf.com/rf_fsk/RFM22B.htm)
-Buck power supply gives 5 and 3.3V rails from a 8-35VDC supply
There's pads to tap the serial off the LEA and run it into a serial-usb adapter. Requires decent soldering skills to assemble. reply here if you're interested in the gerbers or Eagle files.
Will u please give me some sample of s1315f raw output in stq binary?
Here you go... collected with a survey grade antenna:
I am also working on a project with the ublox LEA-4T dev kits and RTKLib. I tried a while ago to get an RTK solution using RTKLib but the quality was quite poor, as the solution drifted by at least 10m to 20m. This was a static test using two antenna's supplied with the ublox recievers (http://www.u-blox.com/en/gps-antennas/gps-antennas/ann-ms.html), mounted on tripods about 2m apart in an empty parking lot (reasonable sky view, I would say). Does anyone have any advice as to how to improve this solution? Perhaps better antennas? What are some good (reasonably priced) options? I do intend to have them on some flying vehicles eventually, so the smaller the better!
Something sounds wrong with your system. What accuracy/variation do you get if you process the individual data sets on their own as 'stand alone'. Average the position of the 'base' and use this as the 'seed' position for the RTK/PostProcessing.
It's likely that the antenna is your biggest problem at the moment ("Compact size: 40 x 48 x 13 mm" is not much of a ground plan). If you can find a reference station close to you, then you will remove the error associated with the 'base' of the system for now and confirm the rest of the system is working.
Drop me a PM if you want some sample data to play with.
Thanks for the reply, I tried exactly that - averaging the single solution of the "base station" and using that as the initial position of the base station for the RTK solution, though from what i've read an error of a few meters on the base station should not affect the relative positioning greatly, is that correct?
I agree, I think the antenna could be a large issue, do you have any suggestions of some relativley small and inexpensive antennas to try out? I assume the reference station you are talking about is using the NTRIP base stations? I am in Waterloo, Ontario, Canada and I could not find anything close by which streams the data for free. Any suggestions?
'About right' for the base station should mean that the RTK offset position is good.
No real suggestions for antennas, but here are a couple of reference stations moderately close (~50km might be pushing it).