I know that it is taken as common knowledge that a hexacopter with a single motor failure cannot be flown beyond a controlled crash landing, but this is just plain wrong with our ArduPilot Mega hardware and if we can't do it, it is because of inadequate software.

I have a KK X525 quadcopter which I have flown very successfully for some time with the APM2.

One ESC on it is now progressively failing, continuously reducing power to it's motor after about 2 minutes flight, but all you notice immediately is the need to feed in a little more throttle and eventually it starts descending even at high throttle.

Once it touches the ground you need to switch out of any auto mode to prevent it flipping over (usually too late).

But the moral is that even with one motor failing it tries to maintain level and flight is still controllable.

On a hex if one motor fails all you should have to do is shut off the opposite motor and it should still be completely controllable as long as you have sufficient thrust in the remaining 4 engines to hover the copter.

Auto shutdown of an opposite motor could be handled entirely in software on the APM and control should be fully maintained. as you still have 2 motors turning each way and the lefit points are still symmetrical.

No spinning, no loss of control, no nothing.

If the Arducopter doesn't do this it should and I would definitely like to know what you think.

This could be a tremendous safety and long term cost advantage for a hex and not require an Octo as previously thought.

Views: 4637

Tags: Failure, Hexacopter, Motor, Octocopter, Safety

Comment by Rob_Lefebvre on November 15, 2012 at 7:51pm

When your ESC slows. does it completely stop?  There's a big difference there.

Fact is, while it is technically possible to fly a hex with a motor out, it is VERY difficult.  It's also true that while flying an Octo with a motor out is possible, this also is not as easy as it seems.  Many Octo builds cannot fly with a motor out, and the reason is usually due to not having enough reserve capacity for the given payload.

I have flown my DwarfStar Octocopter with as many as 3 motors stopped. This is, as you say, a controlled crash.  2 motors stopped is not bad, just a little wobbly.  And 1 out, well, you almost wouldn't know.

Comment by Kevin Brown on November 15, 2012 at 8:24pm

When I lost a prop on my Y6 hex it cam out of the sky like a rock........no chance. A regular hex might have had a better chance I will see My Y6 is becoming an HK HAL hex

Wiki Ninja
Comment by Gary McCray on November 15, 2012 at 8:35pm

Hi Robert, I agree that with an ordinary Hexacopter without APM control you are right, BUT!

Actually the point is, that if you have sufficient power for 4 motors to provide lift to the Hexacopter, with a properly programmed APM board it should not only be possible to control a Hexacopter, it should be virtually identical to normal copter control.

The 2 opposite motors on a hex spin in opposite directions, so if you kill the motor opposite the one that failed, the Hex is now a fully symmetrical quad copter and can theoretically be controlled just like a plain old ordinary quad.

And with the APM's built in Gyro and accellerometer based stabilization it should be a breeze to provide total control.

In fact, if the APM isn't doing that it is entirely a matter of providing the programming to cause it to do so, all the stuff you need to perform this transition is built in.

If you took a stock APM based Hex and programmed it for a Quad frame and disconnected the two (opposite) and not interfaced motors it would fly just like a Quad.

It is in fact a perfectly adequately symmetrical Quad.

APM programming should be all that is needed and it would be a hell of a safety feature.

This would necessitate that you have sufficient reserve capacity to actually fly on just 4 of the 6 motors, but my quad easily has a 3 to one thrust to weight ratio and that is pretty easy to build in with all our modern brushless and lithium stuff, unless you are trying to lift a case of beer.

Comment by Jonathan Price on November 15, 2012 at 9:28pm

Wiki Ninja
Comment by Gary McCray on November 15, 2012 at 9:32pm

You would need to treat yaw slightly differently than for a true normal quad since the 2 clockwise props would be next to each other as would the 2 counterclockwise ones. They are opposite each other on a normal quad. But it should be fully flyable and the normal auto stabilize control on the APM could easily compensate.

Wiki Ninja
Comment by Gary McCray on November 15, 2012 at 9:40pm

Hi Jonathan excellent video.

It certainly shows the capability of auto stabilization like we have built in to APM.

Certainly seems like it could failsafe a hex against any single motor failure with a little work.

The Germans always seem to get there first.

Comment by Andrew Chapman on November 15, 2012 at 11:02pm

I was wondering about a similar emergency failsafe mode for quads, whereby in the event of an engine failure it would spin in yaw to prevent it flipping, and throttle up either to a constant max (not too high as to oppose the yaw spin) or even detect when it is close to impact and use max throttle to slow the fall while it still has momentum spinning it on the yaw axis.

Just thinking to try minimise damage, both to the quad+payload as well as (more importantly) anything or anyone below.

But I wasn't sure whether the APM could even detect an engine failure in all cases - can it?

3D Robotics
Comment by Chris Anderson on November 16, 2012 at 12:20am

What you need is a closed-loop motor controller, where the APM knows that a motor isn't spinning. We're working on that and should have it out in a month or two. 

Comment by Jonathan Price on November 16, 2012 at 1:05am

I think closed-loop motor control is a good idea. There are already open source firmwares for ESCs with Atmega chips. I imagine it would be fairly simple to rework SimonK or another firmware to replace the PWM input signal with bidirectional I2C, CAN, or SPI. That would provide the APM with ESC and motor health data, as well as some other potentially useful stuff such as RPM, current draw, battery voltage, and temperature.

However, I think this kind of failsafe could be implemented without that kind of feedback, and maybe should be implemented without it, since some failure events such as prop break or partial motor dismount wouldn't be apparent to the ESC.

To stop some motors completely while sending others to 100% while in level flight would require that the APM be allowed to command 100% yaw, pitch, and/or roll, but I think APM has limits that prevent that. Is APM stable without those limits?

Comment by Rob_Lefebvre on November 16, 2012 at 8:09am

Gary, you touched on the problems with a hexa, that is that when a motor goes out, and you shut down the opposite, you end up with a quad, but that quad is not balanced well.  For example, if you lose Motor 1, and thus shut down Motor 2, you will see that you have a quad with two CCW motors in front, with two CW motors in back.  This is obviously a big problem.  You have a huge pitch-to-yaw coupling, and yaw-to-pitch.  Basically, you would have to throw yaw control out the window.  This presents a problem for most pilots.  The only hope would be to kick into Simple mode automatically, but we would need to notify the pilot that this is happening.

No matter how smart the ESC's are, if they have feedback, failure detection etc. you can't get past the fact that the airframe becomes aerodynamically unstable.  There is this pitch-yaw coupling.  Yaw control will have to be discarded, and the pilot will thus lose orientation.  The only hope would be to have an immediate forced landing.

Asctec has worked on this problem, what they do is actually spin the opposite motor *backward* when needed to maintain yaw control.  But still, it isn't pretty.

An Octocopter is necessary to provide really good failure protection.  A Decacopter would be even better.  All you would need to do is shut down the opposite motor (which conveniently rotating counter to the first), and still have it be aerodynamically balanced.


You need to be a member of DIY Drones to add comments!

Join DIY Drones

© 2014   Created by Chris Anderson.   Powered by

Badges  |  Report an Issue  |  Terms of Service