Inverted Flight

For Copter 3.6 an inverted flight function was introduced for heli. This was originally put in the code for the collective pitch quadcopter frame, which actually is not a helicopter. It shares no common traits with heli’s - the only reason this frame ended up in helicopter is because it uses the Heli RSC system to control its throttle.

At the time it must’ve looked like a good idea to add inverted flight for helicopters too. So it was put in and only ever tested in the Sim. This was done before Bill and I took over heli. When I learned about it I was probably the first one to actually test it in a real helicopter. It was found to not work correctly.

It is a function that can be assigned to the function switch in your transmitter. You can flip the switch and it rolls the helicopter inverted. When it was found to not work right I threatened to just remove it from the code. As we already have acro and you can fly inverted all you want and everything works correctly in acro. Then we forgot about it and moved on to other projects.

Copter 3.6 rc1 then arrived on the scene for testing and what do we have for new features in heli? Oh yeah - inverted flight. What do we do with that? Bill and I are not comfortable with having something in the code that is known to not work right. So I put together a branch to remove it. But this controversial because it’s a new feature, despite the fact it don’t work right. So Bill came up with a more elegant solution to leave in place for the CP quad and disable it for heli. The PR that disables it is here, with the discussion on it.

We are inquiring about interest from our users in this feature. Since it does not work properly we refuse to have it in the mainstream code. And we feel it is unneeded in heli anyway. But if somebody wants to play with it and develop autonomous aerobatic features we will create a special ArduHeli dev branch for it and provide a build with it enabled. So users can experiment with it if they want. If there is no interest it is a waste of time and we have much more important things to work on in heli.

So please voice your opinion. If there is even one user interested, it is easy to reverse the code in the PR and do your own build with it enabled. Or if you do not know how to do that, we will provide a special build for you to experiment with it.


I personally find autonomous acrobatic features dangerous. It will be tried by pilots who cannot fly an Helicopter in Acro manually. If you don,t know how that feels you will do wrong decision. Like to low altitude or to close to people or objects. And when the Helicopter is of a good size it will go wrong without experience to fly with a massive bang. The low Quality cheap Helicopters may come back with those features with automatic 3D. The first cheap drones like the Nova from HobbyKing have Ardupilot FW as a sample for my argument. Nothing wrong with the Nova I have one too. But automatic 3D Heli is a different story. For me it is a no, no.
People might get killed by those long blades.

Yes, we’re revising this and rolling back a bit. I only made the first post because it was thought to be a valuable new feature. Bill and I learned of another experienced pilot, who happens to be a several-time F3C national champion, that tested this feature with a heli (before we disabled it) and ended up in a crash.

We have no logs from that crash either because of another undocumented feature. The SD card has an automatic ejection system to save itself in a crash. But it is not equipped with a tiny parachute so you can’t really see where it went after it ejects from the exploding wreckage.

This pilot experienced total loss of control after engaging inverted flight. So it will remain permanently disabled in heli and we will not support even testing it.

I have a theory as to why we are seeing loss of control in the real thing with this.

This happened to me a few days ago testing a Loiter landing and was caused by a simple GPS glitch. You can see the heli react like somebody kicked it when the EKF went bonkers. You’ll have to excuse my lovely bride’s comments as she is also an experienced pilot (fixed wing) and couldn’t believe what she was seeing. When it tried to go into dynamic rollover I had full left aileron applied and it refused to respond.

Unfortunately, after further testing and figuring out how to induce a GPS glitch I found this can happen in Stabilize or Acro as well, as the GPS is fused with the attitude solution even in the non-GPS modes. If the attitude solution leaves the scene even the most experienced pilots cannot control the aircraft, unless you have a flybar with flybar mode turned on. Fortunately, I do have flybar mode turned on in my gassers. My RC radio uses the flight mode switch to switch in my own custom rates and expo when I go to Acro so it is direct-thru control. And that’s how I was able to save it from the autopilot antics attempting to wad it up. In my electrics I use normal acro as they are more “glitchy” and not near as smooth and stable as the gassers. All the magnetic fields and “dirty power” in electric aircraft do weird things to servos.

As long as the GPS receiver always has a clear view of the satellite constellation the system works. Multi-pathing of GPS can cause the pilot to have a very interesting time attempting to regain control of his/her aircraft in any flight mode. Obviously, in inverted flight the GPS does not have a clear view of the sky. When the EKF starts playing musical chairs the attitude solution between the processes is never the same, and sometimes radically different. In flight a very experienced pilot can figure it out in Acro with straight-thru control - as long as you’re not doing something ridiculous like playing with inverted flight functions. Close to the ground it is very very difficult, if not impossible, unless you’re really fast on the flight mode switch to disengage the autopilot.

EKF “lane switching” was traditionally blamed on vibration or IMU aliasing. But my Raptor has vibes less than 6-8 and the IMU’s are perfect - and yet a simple GPS glitch can cause the above. IMO, GPS is not anywhere close to being reliable enough to fuse it into the attitude solution in any flight mode. It should be used for navigation only. But that is not our department.

Bill and I talked, maybe six months ago or so, about creating a gyro-only Acro flight mode for heli’s so heli pilots have a true acro mode, and a bailout if this system goes on vacation. Bill is really good with the attitude controller and with Leonard’s help we may still be able to do something like this. Which would be totally awesome for experienced pilots as then aerobatics could be done without fear of blowing the EKF. And if the system blew in flight you’d still be able to fly your heli as if nothing happened, with gyro assist like an old tail lock gyro on a flybar. That would be just way cool, as I’m pretty sure that would open up ArduPilot to being able to handle 3D flight as long as the gyros can handle the G-forces and angular rates a full-blown 3D heli can produce. When it comes to features, that would be one worth working on IMO.

1 Like

All the magnetic fields and “dirty power” in electric aircraft do weird things to servos.

Chris, I have a separate ground wire from the main motor block to the batteries - minus, through the tail tube to the rear motor block, all grounded. It seams to work for me!