Group flights of drones with RTK GPS and real-life precision of RTK receiver as main navigation aid


Lately we have been experimenting a lot with drones equipped with RTK GNSS receivers.

In our setup we used North Surveying RTKite receivers together with Pixhawk flight controllers.

This is a short video with fully automatic flight of 2 drones using RTK GPS. Initial distance between drones was 2 meters, vertical separation 2 meters:

We also conducted a lot of flights to test real-life precision of RTK-equipped drones.

Our test scenario was pretty simple - in take-off in automatic mode, horizontal movement of 10 meters, return to home and land.

We have compiled a table of measurements which can be found below.

It is pretty clear that in RTK Lock mode drones have very high precision. The precision can be even better – we didn’t tune PIDs and the wind was quite high, so the drones had a little drift just before touchdown.

With the RTK base station switched off, drones had precision similar to usual GNSS receivers, perhaps a little bit better because RTKite is  L1/L2 receiver.

Test

Ublox GPS receiver

RTK Copter 1 w/o base

RTK Copter 1

RTK Copter 2

RTK Copter 3

1

1.20

2.00

0.15

0.3

0.3

2

4.00

0.80

0.15

0.3

0.4

3

3.50

1.50

0.10

0.4

0.05

4

2.40

1.00

0.05

0.3

0.4

5

2.90

0.90

0.15

0.3

0.3

6

0.90

1.50

0.10

0.25

0.3

7

1.50

1.70

0.10

0.4

0.35

8

2.40

1.40

0.60

0.3

0.15

9

2.10

1.70

0.20

0.7

0.2

10

2.00

1.00

0.20

0.2

0.15

11

0.50

1.20

0.00

0.3

0.25

12

1.00

1.40

0.05

0.4

0.4

13

0.50

1.50

0.15

0.4

0.3

14

0.40

1.80

0.20

0.45

0.4

15

1.50

1.30

0.15

0.4

0.2

16

0.70

2.10

0.30

0.5

0.15

17

2.00

1.70

0.15

0.45

0.15

18

1.80

1.70

0.20

0.25

0.1

19

1.30

2.40

0.05

0.35

0.25

20

1.45

1.80

0.30

0.4

0.15

Max deviation

4.00

2.40

0.60

0.70

0.40

Avg deviation

1.70

1.52

0.17

0.37

0.25

A word of caution for everyone eager to use RTK receiver as the main one connected to the autopilot and used for navigation, not only for geotagging - we found that in situations when RTK receivers loose RTK lock, the autopilot can suddenly get very different coordinates. This may cause the autopilot can send the drone in an unpredictable direction. This is especially dangerous when flying at low altitudes.   

We are currently working on this issue together with North Surveying support to assure smooth transitions between different RTK receiver modes.


For any inquiries regarding the RTK system, feel free to drop us a PM or e-mail us at ugcs@ugcs.com

More information on our ground station software UgCS can be found at www.ugcs.com


Good luck and safe flights,
UgCS Team

Views: 2261

Comment by kiki boy on July 4, 2016 at 1:23am

I think this comment is not needed :

perhaps a little bit better because RTKite is L1/L2 receiver

Comment by Bernardo Espinosa on July 5, 2016 at 1:42am

Thank you for the video Kistaps, looks great! And yes, I think the issue is just a matter of fine tuning and ensure the data link, woking on that.

If you want to know more about the GNSS RTK RTKite receivers, take a look:

http://gnssrtkmodule.com/

http://northsurveying.com/

Cheers!


Developer
Comment by leonardthall on July 13, 2016 at 5:02am

Very nice!!

It is worth keeping in mind that for integration into the EKF it is probably easier to detect and handle an instantaneous change in position than a gradual change in position when you go from RTK to normal GPS.

If there is an obvious change in state that the autopilot can read then the EKF can reset the new position without any glitch taking place.

Comment by Bernardo Espinosa on July 13, 2016 at 4:41pm

Dear Leonard, thank you for the insight. As we are RTK experts but not UAV experts, we have this recurring question with our users. In fact that the GNSS loses and regains RTK solution is not an issue per se, however, as the solution appears to "jump" as it is not averaged or smoothed in any way like the L1 single GPS do, it presents an unpleasant behaviour for RTK only UAVs., as the changes, as you mention are indeed instantaneous instead of gradual, as they do on averaged solutions towards the EKF.

Cheers!

Comment

You need to be a member of DIY Drones to add comments!

Join DIY Drones

© 2020   Created by Chris Anderson.   Powered by

Badges  |  Report an Issue  |  Terms of Service