[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Paparazzi-devel] simultaneously using two autopilot

From: Chris Gough
Subject: Re: [Paparazzi-devel] simultaneously using two autopilot
Date: Thu, 28 Mar 2013 13:04:02 +1100

> My understanding is that Tridge was able to successfully use a Dual APM setup 
> in one
> of the previous years at a UAV competition. Does anyone know any of the 
> detail there?

Yes, that's what I was talking about - Tridge and I am on the same
team (CanberraUAV).

We did have a working fail-over system but abandoned it on the bench.
It basically didn't care what the autopilot was, it just a bunch of
PWM inputs and PWM outputs (plus sniffed the tx line of the serial
link of each autopilot for MafLink heatbeats), with it's own GPS,
barometer etc. It was also the fail-safe device, if one of the
fail-safe rules passed it would set the actuators to 'dethermalise'
(spiral dive) and ignore it's PWM input from the autopilot. It did
it's job (pass scruteneering), but I honestly believe the system would
have been safer without it. I think our eventual design was basically
equivalent to the twin Lisa/M setup that OpenUAS went with (except
with APM's), I'm sure I've seen code related to that somewhere but
couldn't find it just now (can anyone share a link?).

Another team (TCUAV) had what seemed to me like the best fail-safe
design. Their fail-safe device was a fully independent autopilot (not
connected to normal flight control actuatiors). It had it's own engine
kill actuator (fuel cut-off) and parachute deployment actuator (and
also maybe some kind of dethermaliser - airbrake or something, I don't
remember). It just sat there waiting for a fail-safe rule to pass, and
then killed the throttle (and maybe dethermalised into a spiral glide)
and waited for AGL to fall below maximum parachute deployment
altitude... and deployed the parachute. I think it's fundamentally
better than having more complexity 'inline' between the autopilot and
the plant.

(The reason it didn't deploy the chute in the first stage was because
high altitude deployment could result in a long drift, so they set it
up so that maximum possible drift didn't put the vehicle somewhere

So I should qualify my previous statement - I think redundant
fail-over autopilots are difficult to make worthwhile, but that
doesn't apply to independent fail-safe devices.

Redundant/fail-over components within the autopilot (e.g. IMU, GPS)
seems possible but as far as I know that development work hasn't been

Chris Gough

On Thu, Mar 28, 2013 at 9:24 AM, Drone Savant <address@hidden> wrote:
> My understanding is that Tridge was able to successfully use a Dual APM setup 
> in one of the previous years at a UAV competition. Does anyone know any of 
> the detail there?
> -KF
> On Mar 27, 2013, at 5:01 PM, refik <address@hidden> wrote:
>> Hello,
>> Thank you very much for all of your answers. I am convinced to use a single
>> autopilot system. However, there are still some question marks.
>> I left my umarim_lite_v2 board open for 2-3 days. When I came back, the gyro
>> was not working. I am not sure about why it was burned, maybe it was heated
>> a lot and burned, or I burned it with ESD. Anyway, if it burns again in a
>> long duration flight, it will be a very big problem.
>> Autopilots are reliable, but maybe the IMU components are sensitive and
>> fragile. I will test my umarim and IMU's on the ground for several days
>> again. If it burns again, than I will use a different autopilot or IMU.
>> If it is possible, we can connect two IMU's to a single autopilot. Also, we
>> can connect two GPS and two modems.
>> Cheers,
>> Refik
>> ----------------------------------------------------------------------
>> Message: 1
>> Date: Wed, 27 Mar 2013 11:55:17 +1100
>> From: Chris Gough <address@hidden>
>> To: address@hidden
>> Subject: Re: [Paparazzi-devel] simultaneously using two autopilot
>>       systems for reliability
>> Message-ID:
>>       <address@hidden>
>> Content-Type: text/plain; charset=ISO-8859-1
>> Hi Refik
>> In my opinion it's difficult to make a the system more reliable by
>> adding complexity. If you have an additional component choosing which
>> autopilot should be in control, that device has to be more reliable
>> than the autopilots otherwise the system will be less reliable than a
>> single autopilot. The autopilots are very reliable, so it's a hard
>> ask.
>> In the Outback Challenge competition we were required to have an
>> independent failsafe device. Initially developed a "failsafe/mux"
>> device that with a "failover feature", it would try falling back to
>> the a spare autopilot before triggering a failsafe (deliberate crash).
>> We abandoned that because we felt it was less secure than having a
>> simpler failsafe and a single autopilot. To many wires, an immature
>> component on the critical failure path, more complexity than
>> absolutely necessary. I'm not convinced the failsafe made the system
>> any more secure either, but it was necessary because of the rules of
>> the competition.
>> Redundant communication links do make sense if link reliability is
>> important in your application. Any given link can fail for a number of
>> reasons, spatial and spatial diversity of multiple links probably adds
>> more than the additional networking component takes away.
>> For redundant GPS', I suppose the information is there to chose "the
>> best of many" one but my guess is that the benefit would be marginal
>> compared to single, well installed GPS (good location, good cable
>> management).
>> I don't know about redundant IMUs.
>> Split control surfaces (redundant servos) are a common precaution on
>> larger airframes.
>> Chris Gough
>> On Tue, Mar 26, 2013 at 7:21 PM, refik <address@hidden> wrote:
>>> Hello,
>>> In paparazzi, is it possible to use two complete autopilot systems for
>>> reliability ? (each system includes GPS, imu, transmitter and autopilot,
>> if
>>> one of the systems is gone, the system automatically switches to other).
>>> If it is not possible currently, I think that it will be a good choice to
>>> implement.
>>> We will try to fly an aircraft for 24 hours within 20km, therefore we will
>>> need a reliable autopilot. What configuration (autopilot, imu, Gps and
>>> modem) do you suggest to use?
>>> Cheers,
>>> Refik
>>> _______________________________________________
>>> Paparazzi-devel mailing list
>>> address@hidden
>> --
>> .
>> ------------------------------
>> _______________________________________________
>> Paparazzi-devel mailing list
>> address@hidden
>> End of Paparazzi-devel Digest, Vol 108, Issue 77
>> ************************************************
>> _______________________________________________
>> Paparazzi-devel mailing list
>> address@hidden
> _______________________________________________
> Paparazzi-devel mailing list
> address@hidden


reply via email to

[Prev in Thread] Current Thread [Next in Thread]