Hello,
Maybe the idea is crazy but here I go
Is possible (or anybody did it before) to use two Ardupilots working together?
One master and the other slave. If the master don´t works (instant) a transistor conection wake up the slave Ardupilot and it will take the control of the drone. The second unit is connect to a external safe battery and It can swap the control to Master unit or Slave makes a mandatory return to home to landing or continue working. This reduce much the crash probability, don´t you?
Excuse me for my poor english.
Thanks
Replies
Where this idea left? Someone knows how to run a doble or triple flight controller?
What people originally regard as crazy sometimes turns out to be genius. In your case, sorry, but its not even remotely crazy. :) (See what I did there?) :)
Anyways,
I was thinking about this about an hour ago and am happy to now be able to share those thoughts here.
While having 2 full fledged $200 ardupilots is a noble idea, it's too much. All you would need is a copilot in the form of a second PCB. This can be handled 1 of 2 ways.
#1 - Tapped in to primary/ passthrough. The board that I am describing connects to the outputs of the APM and then the ESCs connect to that board, it acts as a passthrough. It monitors the signal patterns on each channel, or APM can just say "Still conscious" to it on a separate channel every 100ms or so. Either way, if a lack of updates is detected, boom - the slave board needs to have some sort of SPDT (single pole (signal to ESC) double throw (switch between APM and slave)) to switch the ESC control wires over to slave where it will level out the drone.
#2 - Slave, controls the motors 24/7, takes constant directions from APM, like what you were saying. After having written #1, I think this is better. Every <insert time interval here> APM says "Still conscious, here's the motor speed update String". If APM does not say anything for 2* <interval> Slave starts telling motors something else, probably in an attempt to level out the system. The anticipated problem is update interval, since there will be a [small but NOT negligible] delay in the path of APM -> Slave aka 24/7 motor control board --> ESCs...
Either way, the slave board should have a set of accelerometer/ gyro (as redundant) as well as that other nice stuff on the APM such as GPS, compass, etc... Maybe even a 2nd xBee/ 3DR radio. Emergency siren light, those tracker tags they put on dogs so that you can pinpoint the exact location of landing since GPS isn't accurate enough for that, etc... However far you want to take this...
As far as why not to use a 2nd APM as is suggested here... I have no doubt that the APM is a good device, I'm just saying that is doesn't seem to follow the principle of redundancy. IF the primary APM fails, I certainly hope its a software bug and not a hardware issue, but in the event that it is a hardware issue (perhaps from a bad batch, and you purchased 2 that were made one right after the other (extremely worst case scenario here)) you would not be in good shape. Redundancy means different hardware and different software, at least to me.
I think this is a legitimate question. We can build redundancy into as much of our other gear as we like, but we still have only one flight controller. If we could make a system where there were two (or more) flight controllers running in parallel that would be great. Military spec flight controllers can have triple redundancy on the flight controllers but at the cost of $20k or so. Having multiple APM2.5's seems cheap to me,