Blogs

Lately, a lot of companies have commercialized their own RTK GPS receiver. On this blog I will provide information about accurate landings based on an AsteRX-m UAS receiver which works on L1 and L2 frequency.

I’ve tested landings with an f450 DJI frame with the following setup:

• 11 inch props
• 5000 mAh battery
• Total weight: 2kg
• Autopilot: Pixhawk
• GCS: Mission Planner
• RTK GPS Asterx-m UAS
• RTK correction data: RTCM3
• GPS Antenna: AA_ANT_NM_DF_L (AT1675-504)

Because I also wanted to analyze if the ground effect has influence on the accuracy from the landing I performed 50 landings on a solid platform and 50 landings on a wire frame (shown in the picture below). Before the test I expected that the accuracy on a wire frame would be better.

To improve the accuracy from the landings I tuned the following parameters from the Extended Kalman Filter (we want to rely more on the RTK GPS because it is much more accurate than the standard GPS module) :

• EKF_POSNE_NOISE: 0,1
• EKF_VELNE_NOISE: 0,05
• EKF_ALT_SOURCE: 2 (you can use the RTK GPS in combination with the barometer to measure the altitude)

The following mission script was used to realize the autonomous landings:

1)      Take off from platform to 5m altitude (solid/wire frame)

2)      Fly to waypoint several meters from the platform

3)      Return to waypoint 5m above landing platform

4)      Delay for 2s in an acceptance radius from 10cm

5)      Land on the platform 

Landing results:

The accuracy of all 50 landings was better than 20cm (wind speed was low 10 km/h so there were no outliers). The standard deviation is 4,12cm and the mean accuracy 7,43 cm.

Almost all landings had an accuracy of at least 20cm. I had one outlier of 24 cm because there was a wind burst. The mean wind speed during the test was 20 km/h. The standard deviation is 4,79 cm and the mean accuracy 8,08 cm. The accuracy is less than the accuracy on a solid frame because there was more wind.

Conclusion:

• The ground effect didn’t had influence on the accuracy from the landings (probably because the weight of our drone is quite large, I expect more influence on a frame with lower weight)
• Improvement of the landing results when there is more wind can be realized by aligning the vehicle at an altitude of 1m and realizing a fast descent when the vehicle is on the xy position from the landing spot.
• An RTK GPS shows its potential to land on a base station in open areas or on top of buildings. In more closed environments it takes longer to have RTK fix.

Further improvements:

• What is the accuracy from the landing when there is a lot of wind?
• Is the RTK GPS also accurate at high speeds?
• Is this accuracy also possible with the cheaper L1 RTK GPS? (Let me know if you are interested in this)

An example from an RTK based mission:

https://www.youtube.com/watch?v=m9KKw6Y5jCs

More info: stijn.dbr@gmail.com (let me know if you want access to the raw data and logs from the test)