What happens in a failure scenario with the variuos different FC boards and Software?

[roycruse]

Presumably there is no way the FC can know that a motor has failed and therefore is reacting to abnormal rotational and acceleration data.

So assuming a rig with some redundancy ie Octo or X8 or Y6 etc what would happen if the thrust left after a motor failure was not sufficient in one corner of the airframe to maintain level flight - would the software increase power where it could in that corner of the rig and then if its not enough just end up shooting off in one direction or flipping resulting in a total loss - or would it recognise that it could not maintain an upright position and altitude and then lower the power to the opposite corner and maintain a level but descending flight.

Just curios ?

Obviously neither is ideal and I guess a redundant setup with not enough power to maintain a level hover would be a bad design but hypothetically what would happen with the various different FCs.

 

[Bartman]

here's a starter in support of redundancy and having lots of motors.........http://www.multirotorforums.com/entry.php?2-MK-X8-losses-motor-over-jungle-and-survives

i think in most cases if a copter loses a motor the flight controller will continue to try to make a correction up to the limits of its control authority and you can help that by putting in corrective control commands which would override and correct the FC's input depending on the planform of the kopter. a quad would have to go to zero thrust on the motor opposite the failed motor and even then it would have to be perfectly balanced, all this being unlikely which is why quads do the old flip-and-fall when they lose a motor or prop. other low-motor-count formats have similar results with motor failures. coaxial tri-s and quad's have better survivability but it depends on just how many motors are there, which one fails, how much weight is being carried, weather/winds, pilot skill, etc.
Mikrokopters use a single motor control stream (I2c) with the ESC's numbered to know which commands in the stream are meant for which controller. It's fast and effective but a single ESC failure can corrupt the whole stream taking out the rest of the motors. In Shawn's case his BL failed and the kopter kept flying so maybe the single BL failure theory isn't too accurate after all. who knows?!?! There's a small board available that splits the single stream into discrete command streams for each ESC thus keeping the whole I2c system from crashing if a single ESC tries to corrupt the stream.
Non I2c control schemes (PWM) don't have this problem which is most all of the other controllers which use individual connections to each of the ESC's.
There is an I2c to PWM converter and it works well enough for entry level and hobby flying but it doesn't control motors quickly enough to really match what a full Mikrokopter control system can do.
Bart

 

[roycruse]

Yes Ive watched that video - very impressive - in fact i think it was that video that convinced me that when i build my rig it will be with redundancy in it.

But my original question was in simpler terms - how do the different software packages in the different products deal with a motor out situation.

Do they give priority to Attitude or Altitude holding - ie would a controller realise that even at full compensation it could not hold level flight and therefore decide that staying level but descending pretty hard would be better than flipping over.

Software design is one of my jobs and was simply curious had this scenario been programmed in. For instance after a FC had increased throttle to 100% in the corner with the failure and was still detecting further falling away in that direction would it then start further reducing the opposite corner and if so would it do so even if it detected altitude could not be maintained?

 

[roycruse]

I guess my question can only be answered by the developers of the products as unless any of you have been in that situation and can described accurately what happened in that second of panic as all your equipment started falling out of the sky.

 

[DucktileMedia]

This reminds how good it felt when I was doing aerial video using RC powered paragliders. If the engine ever cut out it was a nice slow graceful glide back to mother earth. I am still not convinced these heavy lift rigs cant carry the extra payload of a chute. BUt this is a whole other topic.

My experiences with motors out or broken props is the unit just fights its hardest to throttle up the adjacent motors to level it out. But it doesnt seem to be a gentle come down as the payload is so high that it really needs all motors to fly smoothly. This may be diffrent in an octo that has twice the thrust as payload but I think a lot of us are loading these down 60-75%+. Also, if a motor goes out when your batery is not full life that can have a huge influence on the recovery characteristics.

Anyways, today is the end of the world so it really doesnt matter. Go fly your helis full throttle straight up in the air and maybe we will see them in heaven.....

 

[jes1111]

OpenPilot is working on a purpose-designed ESC that will (among other things) feature two-way communication. It will therefore be able to inform the FC that something has gone wrong (since it could also detect a broken prop by the sudden change in load) and the FC could then evaluate how to react. The idea of adaptive response (i.e. predetermined action) has been discussed too.

No launch date for the OP-ESC, of course, but progress is slow at the moment whilst the emphasis is on getting the Pro/INS out.

 

[roycruse]

This reminds how good it felt when I was doing aerial video using RC powered paragliders.

Do you have any photos of those - I used to fly and teach full size paragliding.

 

[Stacky]

Roy I wonder if its a case of the FC is trying to keep a multi stable or level and adjust power outputs to the required motors. If its not getting a response from one motor it may not recognize it but its sensors keep telling it to apply more power to counter a problem and so the other motors related keep revving faster till it has stability again?. the other motors taking over the lost motors load by default rather than any instruction that a motor has failed?

 

[roycruse]

hi stacky - i suspect thats exactly how it works - but i dont know have never seen a design or spec document or any source code for a uav controller.

My curiosity is simply that. I never plan to build a copter that will need such a function.

As a software designer (in a previous life) I always think of scenarios and then wonder if and how a piece of software will allow for them - just how my brain works - it used to be my job to make software break and then spec out how to fix it.

if i was taking a top down approach to designing software for a UAV I would start with a really simple loop.

1. (start)
2. [am i being told to move anywhere?]
3a. <if yes - calculate what accelerometer and gyro values I should expect>
3b. <if no - set expected accelerometer and gyro values to zero>
4. [compare actual and expected accelerometer and gyro values - are they the same?]
5a. <if yes - everything is great - do nothing>
5b. <if no - adjust speed of ESCs to achieve expected values>
6. (go back to start)

Do that a few hundred times a second and you got a flight controller

All the cleaver (fun) stuff happens in steps 3a and 5b

Of course Im not designing any uav software but love to find out what makes things tick.

 

[DucktileMedia]

Do you have any photos of those - I used to fly and teach full size paragliding.

I dont sorry. It was nice with my electric model as I could fly up high, then kill the throttle so the attitude was stable, then just glide down. No power needed. Was very comforting in comparison to having $5k in one of these multis. And hands off they would inherently fly straight and level. but they had to obviously move forward and flew pretty fast even at 4+ meters. I may someday try it again with modern equipment, this was 15 years back.

 

[thepelell]

i had a prop come loose in flight the other day with my AQ50D powered Y6 that was quite heavily loaded, i was above the 50% hover.
The board tried its best to keep the copter level and almost managed, but there wasn't enough power from the other motor on same boom to compensate, and yaw was uncontrollable and it happened so suddenly , all i could do was give it max throttle for a while and then just let it settle in the soft snow.
here is onboard vid, let's see if you guess with motor had loose prop.

 

[Macsgrafs]

You are very lucky with the height. I think front left prop?

Ross

 

[thepelell]

yes exactly, lower front left prop came loose.. had been doing several prop swaps lately and stopped using threadlock. I was lucky.