Newbie simulation question - HITL/SITL for hybrid quad-plane

336 views
Skip to first unread message

stuart....@gmail.com

unread,
Jun 22, 2015, 10:01:02 AM6/22/15
to drones-...@googlegroups.com
Hi all,

I am looking for some advice on the best way to go about modelling a hybrid quad-plane.

I am part a group who is trying to modify a X8 Skywalker to give it VTOL capabilities.  We have made a first prototype by attaching a quad frame beneath the airframe.  You can see the results of the first quad tuning flight here,


Before we start doing transition flights we would like to do some HITL/SITL testing.  I have seen that Tridge has been helping the FireFly tiltrotor group setup up something similar, however, if I understand the discussion correctly they seem to be taking the approach of having two separate models.  Ideally I would like to have a single model that works in both arduplane and arducopter.

My current thinking is:
1. Start with the existing ArduPilot arducopter simluation
2. Modify to our engine/propeller config
3. Add the X8 mass, inertias and aerodynamics
4. See how well it matches our flight test data
5. Use the sim to run the pusher engine open loop to see how the arducopter stabilisation reacts
6. Repeat, but start from plane mode in cruise and add some open loop quad thrust

Any suggestions, guidance or constructive criticism would be appreciated.

Thanks

Stuart

 

Tom Pittenger

unread,
Jun 22, 2015, 5:15:12 PM6/22/15
to drones-discuss

Hi Stuart,

The firefly is a Y6 which is a stand alone product using two hardware instances of arduplane: one copter and one plane. Team Tiltrotor are using that aircraft platform to do what you are taking about doing: a single ardupilot able to handle both modes. I suggest you get in contact with them. Perhaps a collaboration would give a superior result to benefit all parties involved.

-TomP

--
You received this message because you are subscribed to the Google Groups "drones-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to drones-discus...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Randy Mackay

unread,
Jun 23, 2015, 3:09:38 AM6/23/15
to drones-...@googlegroups.com

     Starting with the simulator sounds like a very good idea.  Bound to save tons of time in the longer-term.

 

     Tridge says setting up a new simulator is “really easy” and I suppose you’ve figured out that it’ll involve writing another SITL vehicle class like these:

           https://github.com/diydrones/ardupilot/tree/master/libraries/SITL

 

     We’ve actually got a PR to add a pusher to copter:

          https://github.com/diydrones/ardupilot/pull/1932

 

     Maybe it would be best to join mumble and have a chat about how to proceed.  Probably best to do that in the morning Asia time.

            http://dev.ardupilot.com/wiki/ardupilot-mumble-server/

 

-Randy

George ZP

unread,
Jun 23, 2015, 6:11:03 AM6/23/15
to drones-...@googlegroups.com
Hello Stuart,

I am the author of the last_letter simulator, which is build on ROS, and I'm currently trying to construct a hybrid model for the Firfly Y6 model. This model will be a unified one, for both the airplane and multirotor configuration, with tiltable motors.
This simulator has successfully been run as an external simulator for APM:Plane and is officially supported as one.

What Tridge was talking about was the actual code of APM to be first tested separately in copter and plane configuration, and then during transitions.

The downside is that currently there are some NaN issues that occasionally break the simulation on startup and that the framerate during SITL with APM is quite slow, probably due to the time slip bug which Michael Day has reported.
The bugs are investigated daily, though.

You can ask for more information here or in the last_letter's gitter channel, whatever suits you best.

All the best,
George

stuart....@gmail.com

unread,
Jun 23, 2015, 11:44:40 AM6/23/15
to drones-...@googlegroups.com
Hi Guys,

Thanks for all of your helpful replies.

@Tom I will get in contact with Team Tiltrotor as I would like to have our vehicle working on a single flight controller.  We also have a FireFly kit so might be able to assist them with some test flights.

@Randy I am not sure that my skills are quite up to Tridge's easy level yet, but I will work on it.  The AC_Booste PR looks interesting and might be a good way to quickly get a stabilised hybrid mode controller working, however, I would really like to test it on a Sim first to get a feel for how it will react.

@George Last_letter looks interesting, but with no intent of trolling, are you able to describe what benefits you see for doing the simulation work in Last_letter rather than default ArduPilot SITL (JSBSim)?

Stuart

GeorgeZP

unread,
Jun 23, 2015, 12:02:59 PM6/23/15
to drones-...@googlegroups.com
Dear Stuart,

JSBSim, at least to my knowledge, is only able to simulate fixed wing airplanes and possibly large-scale helicopters.
It cannot do thrust vectoring and, according to my experience from when I tried, you will find it difficult to create a new model, tailor-made to the specifications of your X8.

last_letter provides solutions for this, since thrust vectoring has already been implemented for arbitrarily many rotors and independent input channels. Also, the aircraft definition is relatively transparent. There are not enough tutorials yet on aircraft creation, but I believe I can guide you through e-mail or gitter (chat) if need be.

As I said, some points that last_letter is lacking right now is that:
SITL integration is not yet finished. I need to work alongside with Tridge to make it run smoothly.
There is a NaN exception currently which impedes progress right now.
Naturally, the visuals are lacking, with no real-world around your aircraft nor sounds.
In short, you exchange a more flexible and probably more suitable (but in alpha stage) solution, for a stable but more rigid, and possibly unsuitable one.
If you manage to simulate your hybrid X8 in JSBSim, along with your exact aerodynamic and thrust model, then I'd definitely suggest you went for that solution, as it's working NOW, in contrast to mine which will be working soon(TM).

I'm not trying to advertise my piece of software. I don't make money out of it anyway. I'd just be happy if the fruit of my labour was also useful to other people, and in turn they'd help me debug it and improve it.

Sincerely,
George
--
You received this message because you are subscribed to a topic in the Google Groups "drones-discuss" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/drones-discuss/CW5jPiGLCq4/unsubscribe.
To unsubscribe from this group and all its topics, send an email to drones-discus...@googlegroups.com.
Reply all
Reply to author
Forward
0 new messages