# Low-cost centimeter accurate satellite positioning (GNSS/GPS)

I am looking for a cheapest possible GPS setup with a centimeter precision without much HW hacking. I am not able to produce my PCB or do any soldering (though I would do that if there is no other way) so a kind of a easy-to-assemble setup would be welcome. I know about the $900 Piksi thing but that is still too expensive for me. It seems like cm precision should be possible for much less - like employing a 50 USD raw GPS sensor with an antenna and ordinary PC with RTKLIB software. I am not sure if it is better to use two GPS sensor setup for RTK (one base station and one for rover) or whether I can get the corrective DGPS data elsewhere (my region is Czech Republic - there seems to be national grid here allowing to stream correction data for reasonable cost). My application will be in a passenger car so I will not be limited with power source - no low power needed although that would be nice. I will be using the position readings within OpenCV - so I need to get the data into C/C++ code. The application is data collection so I can use raw GPS post-processing. ## 5 Answers The NS-HP is a$50 RTK-capable GPS receiver providing centimeter-level accuracy:

NS-HP is a high performance GPS receiver capable of 2 / 4 / 5 / 8 / 10 / 20 / 25 / 40 / 50 Hz update rate. At 1Hz update rate, NS-HP can accept RTCM 3.x message type 1002, 1004, 1005, and 1006, or SkyTraq raw measurement data from a base station to perform carrier phase RTK processing, achieving centimeter-level accuracy relative positioning.

One NS-HP is needed if wishing to use with other RTCM 3.x RTK base station within 10Km operating range. Two NS-HP is needed if wishing to setup a local NS-HP RTK base station to work with another NS-HP rover.

You are already mentioning that RTKLIB and a raw sensor would be the solution. Looking at the list of supported protocols in rtklib, there is a mention of the ublox modules. The LEA6T modules seem to fit and are not very expensive. There is also a simple break out board available which should fit your requirements of not creating hardware.

• Thanks for an answer. I edited my question to ask for single- or double-antenna setup options. Could you elaborate on that edit? – Kozuch Jul 28 '14 at 9:17
• This should probably be another question. – Jakob Jul 29 '14 at 6:56

Depending on your actual project, you can choose several options.

DGPS accurate to 0.5 to 2 meters for real time application, meaning get corrected position now. Needs one GPS receiver and one data stream receiver, wireless broadcast and internet/cellularphone, feeding real time correction data (from reference station) to the GPS receiver.

RTK is real time. Same setup as DGPS. Normally, need professional/survey grade (not consumer grade) GPS receiver. The data stream format for RTK is different from DGPS.

Post processing mean getting accurate position afterward. Record data from GPS receiver that send out raw data (only higher grade unit does this), save to file. Back to office, use software to process the data against recorded 'data stream' from a reference station.

You can own your reference station by buying your own reference station receiver. Some countries/companies operate this as a service, named, CORS, continuously operated reference stations, at free or fee-based. Through international co-operation, IGS operates this on international scale. Data accuracy, coverage area, how-to-get-data, fee vary by huge extend. You need to find out what you want, can have and willingness to pay. Some countries offer best money can buy service, free of charge, as public service.

Reach RTK (USD\$570) looks really good as it is, but what looks promising is that there is an IMU on board, which when integrated should enable less than 1cm error (std. dev.) once the INS has converged.

You should also consider that, because you are post-processing your data, to use a cheap GPS and simply smooth the data using an exponential moving average or other smoothing technique. Alternatively if you need a high rate of pose data, add an IMU and use a Kalman Filter variant to fuse the data (much more troublesome).

• Yes, "much more troublesome" is a very wise piece of advice here, Kalman is HARD – smirkingman Jan 21 '18 at 22:24

Ublox sell an RTK solution which provides 2cm accuracy. The evaluation kit is €359 and it works beautifully.

• Can you elaborate on your experience with NEO-M8P? There are not many reviews out there yet. I guess it could be more robust with less bugs than Emlid Reach or Navspark? – Kozuch Oct 21 '17 at 8:09
• I wrote an in-depth review and test calvert.ch/maurice/2017/08/28/… – smirkingman Oct 26 '17 at 18:47