Discussion on RTK GPS Systems

PIX4Dcapture PIX4Dcapture Questions/Troubleshooting
1

Hopefully this thread will allow us to share our experiences with different RTK GPS units and allow feedback & ideas from others.  It will focus on finding cheaper solutions to allow us to collect accurate Ground Control Points.  

 

I personally use a HiPer SR from Topcon as a Network Rover.  It’s L1,L2, & Glonass.

https://www.topconpositioning.com/gnss/integrated-gnss-receivers/hiper-sr

My setup is a $8,500 HiPer SR receiver plus a $1500 Data collector.  I use my IPhone as a WiFi Hotspot to allow my data collector to receive Base Station Corrections from a NTRIP Caster.  This setup is commonly called a “Network Rover” because it’s a RTK Rover using Base Station Corrections sent over the internet.  In most circumstances I can get centimeter positions “real-time” with a proper workflow.

 

The Cheapest Turn-Key Solution for centimeter positions that I know of is to purchase the Dual Frequency X90-OPUS Receiver for $1880 and log static data for 30 minutes to 2 Hours.  (Longer occupation times are usually not required for UAV mapping precisions).  http://x90gps.com/index.htm

You press 1 button to log the Static data which you later upload to the OPUS website (http://www.ngs.noaa.gov/OPUS/index.jsp) and a x,y,z solution is emailed back to you.  

You can use this receiver to determine precise base station coordinates for a DIY RTK system, or simply log static data for each of your UAV Ground Control Points.

 

 OPUS Map of Estimated Precision

 http://geodesy.noaa.gov/OPUSI/Plots/Gmap/OPUSRS_sigmap.shtml

 

 

For the DIY crowd :

What I’m most excited about is the PRECIS L1/L2 GNSS board by Tersus for $800

http://tersus-gnss.myshopify.com/products/precis-b01

[EDIT - Updated 02-02-2016]

Manufacturer now sells a complete RTK kit for $1,999

http://tersus-gnss.myshopify.com/products/rtk-bundle-915mhz 

 User Manual

https://cdn.shopify.com/s/files/1/0928/6900/files/UserManulPrecisV1.pdf?13329347394704599628

 Data Sheet

https://cdn.shopify.com/s/files/1/0928/6900/files/Precis_DatasheetV1.pdf?15639098580914865445

  

My plan is to build a DIY RTN GPS (Real-Time-Network vs. RTK) with the $800 PRECIS board, $20 Bluetooth module, Android Phone as data collector running a free NTRIP Client, and obviously antenna/battery/etc.   The Android Phone will receive free RTK corrections through a free NTRIP Client.   Again, this type of setup is commonly called a “Network Rover” because it’s a RTK Rover using Base Station Corrections sent over the internet w/ RTCM protocol.  This setup should provide the same functionality as my $10,000 GPS & Data Collector and provide real-time centimeter positions.

This unit should function fine with any geodetic antenna such as a Trimble Zephyr that can be found on Ebay for $500.   I’m interested in testing the performance with cheaper $50 -  $100 active antenna’s, but haven’t had the opportunity yet.

Real-Time-Network Base Station LIST for each STATE

http://gpsworld.com/finally-a-list-of-public-rtk-base-stations-in-the-u-s/

For example, my coverage MAP for Mississippi shows the location and current status of all the Base Stations in the free MS Network:

I’m discussing w/ the manufacturer of the PRECIS board - the possibility being able to log Raw Static data in RINEX format in the future.  The ability to log the RAW observables would allow us to perform a Static Survey and upload to OPUS when Cellular Data is not available for a Network RTK Correction, or if there is no Base station close enough in the free Network for a specific project location to provide a fixed solution.

 Another option is to purchase 2 of the PRECIS boards, and have them communicate as Base/Rover with the same $50 3DR telemetry radios we use in our UAV’s.  Of course, you still need to determine an accurate position for the base station.  This can be accomplished by first measuring the base position as a Network Rover, or if no cell coverage is available you would log a static session and upload to OPUS if the RINEX format can be output from the board. 

 Note:  Novatel makes several boards similar to the PRECIS in their OEM 6 Line for around $800 also.

http://www.novatel.com/products/gnss-receivers/oem-receiver-boards/oem6-receivers/

 

Single Frequency RTK :

Piksi -

http://www.swiftnav.com/piksi.html

$1,000 for a L1 Carrier Phase RTK kit.

Using single frequency RTK (L1 Carrier Phase) has many challenges in the field. 

 

Reach RTK  

http://www.emlid.com/shop/reach-rtk-kit/

$570 for a L1 Carrier Phase RTK kit, same challenges as above.

 

NavSpark NS-HP

http://navspark.mybigcommerce.com/ns-hp-rtk-capable-gps-receiver/

 $50 RTK Capable L1 GPS Receiver, same challenges as above

 

NS-HP User Manual

http://navspark.mybigcommerce.com/content/NS-HP-User-Guide.pdf

 

NS-HP Data Sheet

http://navspark.mybigcommerce.com/content/S2525F8-RTK_v3.pdf

 

NS-HP  Applications Sheet

http://navspark.mybigcommerce.com/content/NS-HP_Application.pdf

 

The NavSpark NS-HP RTK appears to be the RTK cheapest solution, but will also be frustrating in the field.   This unit will experience the same hardships as the Piksi & Reach L1 RTK’s in regards to maintaining the Carrier Phase Lock.  I am planning to experiment with the NavSpark NS-HP since it’s much less expensive than the Piksi or Reach.  

 

 

 

2 Likes
2

[reserved]

3

I have followed your fool steps pretty closely and agree with you reference open source projects with the short falls of handling carrier phase data for post-processing.

Your professional survey gear is the best way to go at least for now.

I’ve had inconsistent results using RTKlib and OPUS data in Linux and the work flow is very trouble-sum. I’m sure I caused at least some of the errors with bad procedures.

BTW, have you post videos ref the RTN approach?

4

We run a HiPer SR with a Carlson Data Collector as a Network Rover, which is perfect survey needs!

5

Thanks Arturo!

I still haven’t found a better approach to acquiring rapid, easy and accurate 3D control other than through professional equipment.

The perfect “box” from the open source groups hasn’t happened just yet but there are a few interesting prospect out there! 

6

Hi Ryan Fontaine, Have you used the Tersus Receiver?

7

Raul,  I have not tried the Tersus yet.   I do think it may be the best option for anyone starting out on a budget.  A $1,000 network rover is a great deal if you have access to a Real-Time-Network, which most states in the USA do.  

8

Thanks for your answer Ryan Fontaine, only exists one problem, they haven’t paypal service and I can’t find reference about them, but these GPS system it complies with technical requirements for my agricultural application.

9

Now , what would be the workflow of extracting the information and inserting that into the images or into pix4d to create a accurate point cloud ?

10

Update 7/3/16 :

Tersus has released a RINEX converter at  http://www.tersus-gnss.com/pages/technical-support .

This will allow a user to use a Precis-BX305: Dual Frequency RTK Board to calculate base station coordinates with OPUS (in the USA) with an accuracy less than 1 cm.

In many situations, using 1 receiver with a Network RTK Correction (RTN) is sufficient for GCP’s, but the RINEX converter is another tool for the “toolbox”.  For instance, if you have a site that you will need to setup a base station on numerous occasions, I would log static data for several hours a couple times during different parts of the day.  This will ensure different Satellite Constellation Geometry for the OPUS Solution.  This is overkill for most projects, however. 

A DIY’r could purchase the $800 Precis-BX305, spend $30 on adding a bluetooth module and SD card, $100 for a L1/L2 antenna, and use NTRIP software on Android Tablet to have a Network Rover.  This would provide the same results as Professional Grade Equipment, and also the ability to log static data for an OPUS solution if required.

 Any GPS system can produce inaccurate results.  The key is to occupy each of your GCP’s at-least twice.  It’s “impossible” that the GPS system will report bogus results twice that have the “same” coordinates.  We do this even with Professional Survey Grade Equipment.

 

11

Another low-cost GPS hardware option is the G1-m1 system by Geomatics USA LLC for US$999 (L1-only):

http://www.geomatics.us/products/g1/

http://www.navtechgps.com/g1_m1_gnss_receiver/ 

A nice feature is the support for stop-and-go occupations, not just continuous kinematic or discontinuous static occupations.

 

I’m keeping a listing of related options at: 

https://wiki.openstreetmap.org/wiki/RTKLIB-compatible_GPS_devices#Vendors

May I also invite you to the FOSS-GPS list for open source software and harware GPS-related discussion and support:

https://lists.osgeo.org/mailman/listinfo/foss-gps

 

cheers,

-Felipe.

12

John thanks for the time to put together a great resource. 

Here is DROBIT a new one I just saw come out . This looks very interesting to. However its priced a bit out of our price range at EUR8500 for the base and drone month rover.  But claims no need for any GCPs and so its much easier to use. I am still looking at all the pro’s n con’s and specs and will look at it in more detail how it stack up.  Be interested if anyone has bought one and tried it?

13

Greetings: Been following along with great interest.

Just saw another idea appear as a Kickstarter yesterday. The project is called One Button Survey (OBS). Here’s the link: https://goo.gl/3Kuzjz

First, it isn’t cheap. $3900 US. I read through pages (!) of Q and A from the developers forum page and the concept seems sound.

Like many here I have been on a quest for a solution to increase the accuracy of my UAV work without breaking the bank for “pro level” survey gear.

Would like to hear what others think of this idea. To me it seems sound (although I think the cost is high).

Marshall

14

Marshall,

The One Button Survey (OSB) is interesting, the developers have worked hard to “simplify” the field work.

Being it’s a PPK solution, there are disadvantages to it’s application for aerial survey work in my opinion. 

PPK requires Post Processing your data before the positions are known.  

 

I’ll briefly explain my typical field work to point out the main difference.

I will set GCP targets outside the limits of the required survey, say 5 for instance, plus 1 GCP near the center of the site. 

Using RTN (a network rover), I will collect 2 minutes of RTK data for each GCP target before I fly, and the same for a few natural targets to use as check points.  

Then I will fly the site with the type of mission best suited for the project.

Finally, I will collect another 1-2 minutes of RTK data for each GCP as I walk around the site to pick-up my aerial targets.  

Since a decent amount of time has passed between observations, the 2’nd survey of each GCP is using entirely different Satellite Geometry.  I will immediately see the delta between the 2 observations,  and can decide if that GCP is acceptable to me for this project.  If not, I fix it.  This is how I verify the  accuracy of my GPS Ground Control while I’m at the project site.  I’m not sure how that’s done with PPK. 

During the flight and walking through the site to pick up targets, I usually notice more natural targets to GPS.  I’ll also survey these at the end of the job for additional check points or Control.  

 

15

This subject needs much more attention by industry leads in aerial data collection.

Ryan, along with the rest of us who are try to maximizing the accuracies in the photogrammetric outputs to supplement well established industrial workflows and accuracies, would benefit by offering systems to complement and optimize their photogrammetric workflows. I guess 

With many companies offering low cost, high accuracies solutions, or at least a potential solution to traditional processes are still off in the future. I hope there will be a better solution than a pickup full of survey kit, but today traditional SBAS/RTN kits are the standard, hopefully someday not.

16

Ryan: You and I have pretty much the same workflow. Thanks for the details. I am trying to get away from the second “go around” with our RTK setup. Of course the extra time it takes depends a lot upon the size of area we are doing but it is still slower.

PPK is theoretically more accurate than RTK (hence why it is used for control). I guess the only way to fully verify that would be to do both and compare residuals.

Gary’s point is a good one. I wish that one of the companies, for example DJI, would just ask a real surveyor what is required :slight_smile:

It seems that a lot of effort is being expended to get “accurate” locations of the images at the moment the camera shutter opens but the refresh rates of 5 Hz, 20 Hz, or even 100Hz mean that those positions are interpolated, especially as the UAV is traveling at 5m/s or greater. I see that Datumate’s flight controller (called Datufly) is a ‘stop and go’ (ie fly to waypoint, stop, trigger image, go to next) but it still relies on the UAV’s on board (inaccurate) GPS. So I think, at least in the short term, focusing on using GCP’s and workflows that include establishing GCP’s is the way to go.

 

17

Marshall,

My intention was not meant to question the accuracy of PPK, just the implementation with Aerial Survey work (based on my workflows). I’d certainly be interested in a different process if it saves time.  

 I forgot to mention I use Pix4DTagger and their coded targets, So I have to make a 2’nd loop around the site to pick up my targets. For me, the automatic target recognition for GCP’s is well worth it.  No more manually tagging GCP’s in the images.  

I agree with you about having GCP’s instead of accurate image locations.  The Ground Truth is the only way I know to verify the model.  IF I ever did have aerial RTK GPS, there is still no way to prove the model without taking Ground Shots anyway.    No way I would solely trust the image location based on interpolation of aerial GPS and IMU for my Model’s accuracy.  

Tersus makes a RTK board that support all 3 constellations with 2 antennas for flying.  Maybe both positions could be verified by automatically checking the distance between the 2 antennas on the aircraft, a 1-2 meter baseline.  That may work for a fixed wing UAV one day.  But if the Baseline didn’t check, would you scratch that image?  A couple of misses in a row and the flight was worthless. 

http://www.tersus-gnss.com/collections/gnss-rtk-boards/products/precis-bx316 

 

I’m always interested in hearing how other people perform their work, it’s the quickest way to learn.  

 

18

Hi Ryan,

 

Thank you for the great information and everyone for their input. I have extended my mine planning business to create accurate 3d models to do basic planning and volume calculations and creating contour surfaces.

With all the research that I have done, I can conclude that I have most of the things sorted out, beside getting accurate GCP locations. Have you used the Ternus precis board and would you recommend it for a rtk gps tool with a limited budget?

Would you require internet connection at all times to take gcp readings?

That is the only reason I will use it for is GCP, would the Ternus be sufficient.

Have you seen this before? https://www.propelleraero.com/aeropoints

Great way to be very efficient, however, slightly pricey.

 

Cheers

Ruan

19

I am at the point to decide between the Ternus BX305 or the newer Piksi Multi.  I cant see a great deal of difference in that both are using multiple sat systems and seem to have comparable specs and accuracy. So can anyone can give the pros and cons either of them? At the moment I am leaning more towards the Piksi.

20

Ruan I did have a good look at them and I thought the Aeropoints are way to highly priced for what they do. Its basically just electronic GCPs. and that all you can use them for. Where is for very much less you can now buy what are virtually survey grade GNSS system that have many more uses and more flexibility. Eg mount on UAVs, Rover,s Vehicles and for field surveys etc etc. The inflated price I see is more an issue its made in Australia where typically just about everything is very expensive. (Im also Australian in case I will get some kicking over that comment :slight_smile: )  For us out here working in Asia, it would take us much to long to return such an investment for the small role it plays in the bigger picture . That was just my view on it though.