Pushing a Parrot Disco to the limit


I’ve been doing a lot of flight testing of the Parrot Disco with ArduPilot lately, and had some really fun flights on Sunday with my fellow crazy CanberraUAV pilots after everyone else at the club went home as the wind was too strong. We decided to wait until the wind got to its strongest to see how the 800 gram Disco would cope with really high winds. It did very well! We recorded wind speeds of over 40 knots in the flight and it flew extremely well.
Grant had his Firstar there too (see if you can spot it in the video) and he recorded a ground speed of 155 km/hr downwind. Pretty fast for a foamy!
The video is the first test of the on-board digital video stabilisation on the Disco while flying ArduPilot. This was developed by Julien from Parrot, with some assistance from the new AP_Module system in ArduPilot to feed sensor data to the image stabiliser.
Never let anyone tell you the wind is too strong for a fun flight!

PS: the actual takeoff of the Disco starts around 10 minutes into the video. I haven’t yet implemented a transmitter switch to start/stop video

8 Likes

That is proving to be quite a capable little wing Tridge, any plans to shoehorn a vertical camera in there? Try some mapping.

it actually has a low-res downward facing camera (for optical flow). Not a mapping camera though

I wonder if that space behind the AP would fit a small Canon point and shoot.Lifting it would be another thing of course.

I love the Disco for the fact that it is flying arduplane :slight_smile: Thanks Tridge and good job Parrot!

Awesome stuff! Just bought a Disco and flew it today in high winds too…efortless. My Pixhawk driven TBS Caipirinhia couldnt cope :slight_smile:
Anyone know how to fit another Video Tx and TBS Crossfire to this wing? Really battle with the HD feed to my iphone and lack of compatibility with Fatshark HD goggles.

1 Like

Nice!

Once I get my hands on one I will be making a simple to install 3D Printed/CF quadplane mod for it using our OBC drivetrain parts, that will simply slot in between the fuselage and wings. Works out that it should hover as a QP on 3S at around 20A, which means it would be a contender for 18650GA cells if one runs 2P3S so with 6800mAh capacity (288g), which in turn shouldn’t add that much more weight,but should give it a range boost.

Tridge is the battery close to the COG, and roughly how much does it weigh? If you want I can 3D print the parts required and assemble some quad arms with propulsion parts for you, if you send me some pictures of the wing section/connection alongside a ruler for scale, and a mark for the CoG.

Regards

That would be awesome… Please keep us posted

It weighs 190g, and it is centred about 1.5cm in front of the CoG (roughly, I used fingers as a CoG balance).

That’s very kind! I’ll send you some photos, and if it looks promising I could send you a fuselage and wings to play with. I have a spare one that I pulled to pieces in order to put the CHUCK autopilot in a different plane.
I think the main problem will be the elevon servo. If you try a quad arm setup between the wings then the arm will interfere with the elevon servo.
Here are the photos: http://uav.tridgell.net/Disco/images/
Cheers, Tridge

Not a problem Tridge.
It is a little self serving in that I’m interested in a Disco QP myself! :wink:
But I’m happy to share the results and make the 3D print files/parts, or whole assembly available as a kit, for those that don’t have access to a 3D printer.

From the photos I can see what you mean with the Disco elevon servos and plastic arm. But I think that can be overcome by placing a flat CF spar (15x3mm section) under the fuselage side of the wing connection point (which should miss the servo assembly), and mounting it by 3D printing a 2mm thick L shaped bracket, with all the relevant wing spar penetrations etc. and a slot on the bottom of the L bracket to hold onto the quad arm CF spar. The servo arms should manage to bridge the 2mm extra gap without any modification to the Disco, alternatively creating a recess for the bracket by trimming the wing side foam with a hobby knife will work too.

I’ll do some CAD drawings tonight to demonstrate what I mean. For that could you maybe please redo the photos so that they are as perpendicular as possible, along with a ruler for scale, then I can already make up a fairly accurate CAD model using just the images. It would be good to have a photo of each side of the wing connection as well as from underneath the whole fuselage. Could you also give me the dimensions of the battery and a photo from the top with a ruler as scale?

I haven’t quaded a wing yet, but from the photos some optimisations will need to be done to make it work, preferably without many changes to the Disco itself:

  1. The forward motor prop has to clear the quad arms which should work, as the prop arc and low quad arm mount position should leave enough clearance
  2. The spacing between the quad arms is fairly narrow by putting them on the wing connection point, but we haven’t experienced any roll control issues with our narrow Talon QP mod yet. I think it will be fine as it also weighs 1.2kg less.
  3. The quad props need to clear the fuselage and be distanced symmetrically to the airframe CoG. (Within reason, as I think we can more dynamically add quadplane parts than just squarely on the CoG. For example; pushing the quad CoG forward will give it better wing AoA in hover, if we can offset pitch trim in hover, which should work well with some quad motor forward tilt on the motor mounts themselves, plus maybe even some upwards pitch on the elevons)
  4. We’ll have to investigate the best spot to pull the quad arm cables through the fuselage to connect to the CHUCK spare PWM’s, and make a XT60 adapter for the 4x XT30 quad power etc.
  5. The 3D printed L bracket might need some little extrusions on the bottom so when it lands as a quad it sits fairly straight and gives the quad motors a bit of clearance off the ground.
  6. I’ll need to have a look at how to best configure the 18650GA batteries so they fit in the battery spot, on the same CoG (subject to 3. above), with little to no Disco modification. In the mean time I think the Disco battery should be ok for 20A hover, but maybe use another cheaper, easy to replace 3S battery instead for testing. The battery is quite close to CoG so it shouldn’t be to hard to do.

Funnily enough I was already looking at sourcing some spare Disco airframe parts for a build using a PXH mini, and a PiZero controlling a downwards facing camera and comms. Happy to pay for them and freight, or otherwise maybe we can do a swap, custom built up quad arms for the Disco with custom built 18650GA batteries for a CHUCKless Disco? :wink:

Regards, Sam

Hi Sam,
Parrot should have contacted you by now about your Disco QuadPlane project - looks like you’ll be getting a couple of Discos to play with!
We should probably create a “Disco QuadPlane” discussion on here to discuss the build.
Cheers, Tridge

1 Like

Thank you very much for organizing the Disco’s to develop the quadplane on Tridge!

Much appreciated. It will make it much easier to develop a quadplane conversion kit for the Disco.
I’ve sent them my details and also asked if they might be able to provide some dimensioned drawings to build from.

Making a Disco Quadplane discussion here would be great, and I will try to keep it up to date with progress from my end.

Regards, Sam

Can you tell me exactly what controller you use? What distance can you obtain with that controller?
Also, I did distance tests with the stock firmware and controller. There is a geofence at exactly 3.5 km, which seems to be built-in to the firmware (there is no way to set this, not even in the config files). This is different from the geofence you can enable in the app.
Now my question is: does this “geofence” of 3.5 km also apply with ArduPilot on the disco?

I use a Taranis with a FrSky X6R receiver, with FrSky telemetry disabled (as it interferes with WiFi)

I only fly line of sight. I’d expect it to do about 1.5km if I wanted to though.

no, ArduPilot completely replaces the built-in firmware. It behaves the same as ArduPilot on any hardware, no additional restrictions.

That’s interesting, tridge. Can you then still use the app?

no, you can use any MAVLink GCS though. I use mavproxy.

Are there apps on iOS?
Can you make some screenshots at some point?

I don’t use iOS myself, but I believe there are mavlink GCS implementations for iOS. I’m not the right person to ask about it though, I happily run Linux on my GCS

Disco QP Update:
We received the Disco’s from Parrot and are preparing to do the first Disco quadplane build this weekend, and hopefully QP flight tests early next week. I’ll start a new discussion for the Disco QP build as soon as we have something built.

We have also flown the stock Disco and it flies well but we have yet to try Arduplane. Following that we’ll test the latest QP code, hopefully the CHUCK channel and power train setup is straight forward.

3 Likes

@Tridge or anyone else…

Can anyone post a summation of the state of using a Parrot Disco with Ardupilot? You have to bind your own TX, right. So you could use a long range UHF system and then the issue is basically video range?

It does seem to need 2.4 wifi boosters and better ground antennas (based on my reading the owner forums on RCGroups), but that is not a big deal.

How does flying it out out the box compare to Ardupilot?

Will Ardupilot eventually be usable through the Parrot wifi link?