A PX4 AR Drone is born 2nd post

Welcome back to my build log/experiences with this project. 

After playing around with and figuring out where to put the various bits I went on to add the XBee headers, powercable and 3 pin RX header to the PX4IOAR: 


At this point I will skip further as the assembly is well detailed in pictures on the PX4 website. I would add that getting the spacers in between the PX4FMU and IOAR is not entirely simple without 3 or more hands, but just fiddle in the white spacer first, then push down the screw through the FMU, one by one is probably one of the easier ways to do it. 

Here is what you end up with after assembly: (zippo again for size - the quad is tiny!)


Ok, now we need to get this all fitted into the EPP foam body donor of the AR Drone, including the GPS and the FLOW. 

Bottom view (I know the battery is a bit incorrect, I took only one pic of this)


In the above picture you can see both the cutout for the battery as well as that for the flow unit. I am not 100% sure yet that the sonar will not have issues, but I hope not. Would not want to mount it any lower as the lens will poke out then and strike ground if it comes down hard. 

Topside GPS mount closeup, as you can see I cut out a flat plateau first, and then carved a smaller space in the middle for the cable and the chip etc on the bottom of the module:


It's ALIVE!!! Look at the pretty lights! 


Next up at this point is to test everything together, get the startup script correctly set up etc, plus installing the FrSky receiver when my FrSky DR 4-II arrives. 

E-mail me when people leave their comments –

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

Join diydrones


  • I have strong reasons to suspect that my PX4FMU has some weird manufacturing fault, it never worked as it should even with help from the designer of it and several other people. Sent off an email to inquire about what the RMA procedure would be, no answer yet. So it will likely be quite a long time if I can get the FMU fixed somehow, if not I will probably just park it next to my large collection of things that seemed good but turned out not to be. 

    The PX4Flow does seem to work well, I can get mavlink working on it and everything seems to be checking out fine, but without anything to use it for it's just gathering dust. 

    What I was trying to do was to get it all integrated into a system that would allow a rather untrained remote operator to control a flying vehicle in a reasonably reliable way, mostly for SAR and hazardous environment type applications, which is something I have been trying to get working for a while now. On paper it should be possible but I think the tech is still not there, atleast not on a budged a hobbyist can afford. 

  • Thanks Noth, for your information.

    Btw, hope it does work for you sometime soon, would be eager to hear your results :)

  • In case anyone is curious, this setup never actually flew and has since been disassembled and at least for the moment put on hold. If I get back to it and it actually flies one day I will update here with my experiences.

  • Shyam,
    I run the windows eclipse environment for development. This is the standard setup for PX4FMU, if you go through the default installation instructions, at the end you will have eclipse installed and set up for the PX4FMU and PX4IO firmwares and the latest sources also.

    The instructions are all on the pixhawk site as well as in the FAQs.
  • Hi,

    That is great work. I wanted to know what platform and language do you use for PX4MU? Is it Windows or Linux based development?


  • It looks like it does a twitch/jerk but it is indeed the motor controller internal test. I should have made it clearer that what I was trying to say with that is "everything seems to work" :). I made the comment to show that they do initialize and get power.

    Another interesting effect is the LEDs when running with the px4 they do a different kind of slow blink that looks very much like the lights of a real airplane, with all LEDs it looks quite fancy all in all :-).
  • Developer

    Very nice build log. Note that the motors don't jerk, they run their internal test (as the original AR.Drone does depending on its mode) and you should get them one on at a time, in a fast circle. For readers not owning an AR.Drone: the motors don't spin, they just 'tick' and the blade only moves about 0.5" at the tip of the blade. This is done by the motor controller once it's properly initialized and not PX4.

  • Just got XBee mavlink up and running with QGC, also all LED's work on IOAR, and the motors jerk a bit when it powers up. 

    Still need to do calibration, the accelero thinks it's tilted a fair bit while sitting level on a table :) but that is to be expected. 

    Running the last published build, everything else seems fine other than that QGC says disconnected off/on while the modems clearly are streaming data, not sure why but that is tomorrows worry. 

    Another thing I forgot to add is how the GPS and FLOW are currently attached. The FLOW is stuck into it's pit with the cut off ends off some tiewrap going through the front side to side, you can barely make it out in the last pic. 

    The GPS is actually glued, but using UHU Por which is EPP and removal friendly, so I can relocate it on top of the stack later if I need to. Eyeballing it now it looks like the hole spacing for the FMU is the same as for the GPS, if so I will need some more plastic standoffs and screws to get it up there. 

This reply was deleted.