RTK Loiter / RTL Testing with AC 3.2.1 and NS-HP-5

Holidays Greetings!

There is much interest in seeing demo of our $50 pre-order NS-HP RTK receiver on a quadcopter, so we tested last week on a less windy day (with up to 4.5mph wind). A compass module and 5Hz version of NS-HP RTK receiver replaced the original GPS and compass inside the 3DR Iris. These two videos show the result we obtained with default values without any tuning.

E-mail me when people leave their comments –

You need to be a member of diydrones to add comments!

Join diydrones


  • @Oliver Huang:

           Thanks for you sharing, it's great. But I have a question, when the GPS status changed from RTK fix to float or from float to RTK fix in loiter, the drone may fly away suddenly, did you have similar phenomenon? Thanks


  • Well, let's say that you're trying to fly a course through a bunch of obstacles.  Think Red Bull Air Racing but much tighter.  You survey the location of the obstacles in advance so you know where they are.  Significant jumps in position can mean the difference between threading a needle and crashing into an obstacle.

  • Below image shows static testing, initially in RTK fix state (point at of center scatter plot), intentionally stop sending base data for 90 seconds, it goes from fix to float to DGPS mode, then start sending base data again it gets RTK fix. The total drift in this experiment is around 20cm. 


    If from power on to getting RTK fix going through 3D fix, DGPS fix, and float, the jump could sometimes be more than a meter.

    So is this sufficient ?

    GPS error is usually more than a meter. Regardless how good the RTK receiver is, going from normal 3D fix to next better state there will always be a jump of more than a meter. Some filter could always be implemented to to reduce the jump to any desired distance number; it's just a matter of how quickly to go from a less correct position to a more correct position so to speak.

    I don't quite understand the importance of minimizing this jump issue. Could you explain?

  • Have you tried to quantify jumps in position when the device goes from regular GPS to float (or fixed)?  Are we talking jumps on the order of more than a meter?

  • @Pedram Not sure if the question is about how RTK improves drone application in general. If that’s the question, knowing position of a drone in cm-level accuracy would allow consistent repeatable flight path or landing at a fixed location; might be useful for flying over rows of crop for inspection, more accurate air delivery of items, precision landing, taking camera shots at predefined fixed locations in aerial survey…etc. When there are more drone buzzing over sky years later, would there be lanes in the sky like the roads for cars, with these smaller sized drones not taking wider lane than a larger car? …


    @DRONESPAIN Correction data goes into rover, if wish to monitor NMEA on the ground then it’s out going from rover. If changing rover NMEA output to 57600 from 115200 default, and if the radio can support sufficient data bandwidth in 57600 over both directions, then I guess one can use one radio for sending base data and receiving NMEA, provided radio on the ground is using UART interface so one can wire base TX output to radio RX input, and radio TX output to some UART-to-USB bridge breakout board to connect to laptop. 


    Sorry neither our drone pilot nor I am familiar with this GPS_INJECT_DATA MAVLink message. If the ground station can take base data over UART input, pack the data into MAVLink message to send over radio, and Arducopter software decode the MAVLink message and restore the base data to send over UART to the rover, then I don’t see there is any issue. Maybe others who has tried this before could help answer?

  • Thanks Oliver, information contained in your UserGuide is clear.

    I suppose 3dr radio and rdf900 interfaces seemless with your unit.

    When talking about "monitoring" in page 20 of your guide (pls, number the pages in word in future versions!) you say rover unit outputs NMEA to a radio. Is it supposed to be another radio different from radio used for corrections? That is... must I have 2 radios in base station, one to transmit corrections and one to receive those monitoring data on GNSS view software? And 2 radios in rover, one to receive stream corrections and one for monitoring with provided software?

    For us having 2 radios IS NOT A PROBLEM as they are cheap and light weight.This is only a simple question because I read something similar in this one thread:


    Where they talk about use GPS_INJECT.

    Sorry if my question sounds dumb. Thanks in advance.

  • hi

    we have a set of RTK GPS receiver ( for surveying) include base- river and radio modem to send correction .

    would you please tell me how this module helps us in flight ?

    (i know that single positioning is around 5m but  RTK is around 1-2 cm)

    thank you

  • @Nikola Rabchevsky When losing RTK becoming float, it still maintain good smooth track. In this blog with Dynamic Testing with 4.5Km Baseline on a very heavy rainy day in the lower section, the figure below is mostly float solution and few sporadic RTK fix solution, there is no jumps. Note that it’s still under open sky though.



    NS-HP is not intended for urban canyon, under tree signal blockage applications, only for open sky usage. When it lose RTK fix later changing from float to 3D or DGPS fix, there will be a jump, going from 3D or DGPS to float there will be jump. So long as signal condition is bad, not good enough to enter float solution state, the track will be smooth like a normal GPS receiver.


    @DRONESPAIN Unfortunately the $50 NS-HP is totally not a “rtk ready" GPS like NEO-6P or NEO-M8N, we completely robbed users the fun of complex setup and use of RTKLIB. Upon receiving, it’s an RTK rover; setup only needed to configure into base mode. This user guide has the details: http://navspark.mybigcommerce.com/content/NS-HP-User-Guide.pdf

    Does it still maintain a bit of challenge interesting for the users? :-)


    High Precision — NavSpark
  • Hello everybody and thanks Oliver for answering questions here. Your developments seems very good.
    I suppose this 50$ unit is a "rtk ready" GPS like Nep6P or even M8N...

    Are you planning a "working out of the box" release, like base+rover+how-to? I agree with you, buying radios is not a problem, but as far as I'm concerned, the problem with rtk+UAV technology is the lack of information.

    Nobody wants to spend 500$ in a unit THEN spend hours and hours dealing with it to make it work!!!


  • Thanks, Oliver.  This makes sense.  Have you run any tests that show what happens when you lose RTK completely?  Does the SPP solution have an internal Kalman filter to keep the position stable or does it look like a 2.5 meter cloud of positions?

    I'd also like to see some less-than-ideal-conditions tests such as transitions to and from an urban valley or to and from under trees.

This reply was deleted.