Comparing Precision of Autopilots for Survey Missions - DJI A2

We have reached the fourth week of our autopilot precision series. In this series of articles / videos we test the different drones by flying them on a standard photogrammetry route created in UgCS.

So far we have analysed DJi Phantom 4, DJI Naza-M V2 and Mikrokopter Quad XL. In this week’s video we take a look at the results of DJI A2. The particular A2 was set up on a DJI S900 platform and we flew it three times, each time with a different turn type – stop and turn, bank turn and adaptive bank turn.

Here are the results:

Download the KML files and take a look at them yourself in Google Earth:

Get the newest version of UgCS here:

Safe flights,
UgCS Team

Views: 803

Comment by Andrew Zaborowski on April 10, 2017 at 5:12pm

A little short on hard data.  I'd expect some overall error estimate like least square or surface area between actual and desired path, explanation of how the actual path was obtained to exclude GPS error, and a comparison of these results to the already-reviewed autopilots.  All in all it's a difficult parameter to compare without someone claiming that the airframe could be causing oscillations, navigation PIDs weren't optimal, etc.  Looking forward to see how the opensource autopilots stack up to one another.

Comment by Alexey Dobrovolskiy on April 12, 2017 at 4:28am

Results for 3DR Iris with PX4 firmware will be published tomorrow.

Comment by Alexey Dobrovolskiy on April 12, 2017 at 4:39am

Regarding interpretation of our data. It's not a scientific work and we are not pretend to be absolutely true. We would like to give common picture of different autopilots behavior in automatic missions.
We don't use any means to track actual trajectory with hi precision, like additional independed RTK GPS. All data was recorded using standard telemetry from vehicles. It's main reason why we name our cycle "Autopilot precision". Real vehicle precision will include GPS/altimeter error and can be much worse.

Comment by Alexey Dobrovolskiy on April 12, 2017 at 4:41am

There is a mistake in text, right link to data for A2 is

Comment by Rob_Lefebvre on April 12, 2017 at 5:55am

To be clear... you have loaded PX4-Stack onto an Iris?  Iris by default comes with Ardupilot firmware.

If you did load PX4 onto it, you must be intentionally keeping us on the edge of our seats waiting for the Ardupilot test.

Here are typical results.  There are actually 5 laps of this figure-8 pattern overlaid onto eachother.  And this was a 1kg helicopter running at a target speed of 40 m/s and accelerations of 1/2 G.

Here's what that run looks like (this was an aborted attempt).

Comment by Gary Mortimer on April 12, 2017 at 11:05am

History channel stuff there Rob!

Comment by Rob_Lefebvre on April 12, 2017 at 11:49am

Never did get my prize Solo RevB GPS.  Had to buy an mRo. Probably for the better anyway.  ;)

Comment by Alexey Dobrovolskiy on April 12, 2017 at 2:04pm

yes, we loaded PX4 stack on to Iris.

Comment by Kirill on April 14, 2017 at 5:07am

If someone wants to have better flight precision compared to the ones above - take a look at SmartAP (


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

Join DIY Drones

Season Two of the Trust Time Trial (T3) Contest 
A list of all T3 contests is here. The current round, the Vertical Horizontal one, is here

© 2020   Created by Chris Anderson.   Powered by

Badges  |  Report an Issue  |  Terms of Service