Restructure INDI to limit estimated disturbances

[EwoudSmeur]

I restructured the INDI controller, keeping the workings exactly the same. The benefit of the new structure is that we replace the increments by a disturbance estimate, such that in the control allocation we allocate the complete control objective (specific angular accelerations and thrust), not increments. This allows us to limit the specific angular accelerations (read: moments) that we allow the drone to allocate.

As an example, the rotating wing drone has a tail, which will automatically reduce the sideslip. We don't want the drone to stick to a specific heading, and in doing that 'fight' with the tail. With this update, we can limit the estimated disturbance such that we only control yaw to a limited extent.

What is also included is a feed forward on the yaw rate from RC if the yaw attitude gain is 0, otherwise it would not be possible to control yaw anymore with the sticks if you want to fly without a fixed yaw attitude setpoint.

[gautierhattenberger]

This modification is currently not fully compatible with #3248. Not on the principal, but in its implementation, since the other PR is completely changing the workflow for stabilization and RC inputs. Would be possible to have a discussion (online if possible) around the strategy to merge some of the awaiting PR that can have an impact on stabilization ?

[EwoudSmeur]

Hi @gautierhattenberger, the impact of this particular PR should be minimal, at least that is the idea, but I'm happy to discuss how to best approach merging the PRs. I can do today in the afternoon, Friday afternoon, and Monday, just let me know when you prefer!

[gautierhattenberger]

Thanks, replied by email

[EwoudSmeur]

Now without the yaw stuff!