Copter: simple mode fix

[rmackay9]

This resolves issue #33936

Replaces update_simple_mode with apply_simple_mode that takes roll and pitch input arguments. The rotation is now done within the various get-pilot-desired-xxx methods removing the need for each flight mode to call update_simple_mode and also removing the slightly ugly call to overwrite RC_Channels' control inputs from the flight code (e.g. no more calls to set_control_in)

This fix also introduces these changes which I think are OK:

• Drift mode and Precision Landing's repositioning now support simple mode
• Save trim also works in simple mode although the user needs to be careful not to change the vehicle's heading while landing and saving the trim.

This has been tested in SITL in Acro, AltHold, Drift, Loiter and Land modes to ensure both simple mode and super simple mode work as expected. Below are the modes I've tested:

• Acro (both multicopters and heli)
• Drift
• Sport
• Precision Landing repositioning
• Save Trim

A new autotest borrowed from PR #33939 ensure the behaviour is correct so that this doesn't happen again

Below are some screen shots of testing save trim in SITL. During this test did this:

• manually set AHRS_TRIM_X to 0.0872 (e.g. 5 deg)
• armed and took off to 20-ish meters
• engaged simple mode
• rotated the vehicle's yaw to point south
• because of the AHRS_TRIM_X the vehicle drifted quite quickly to its right
• applied RC roll and pitch input until the vehicle came to a stop
• manually landed and used the aux switch to save trim
• checked the AHRS_TRIM_X, Y values to confirm they were close to zero (much improved from the initial 5deg)

Classification & Testing (check all that apply and add your own)

• Checked by a human programmer
• Non-functional change
• No-binary change
• Infrastructure change (e.g. unit tests, helper scripts)
• Automated test(s) verify changes (e.g. unit test, autotest)
• Tested manually, description below (e.g. SITL)
• Tested on hardware
• Logs attached
• Logs available on request

[tridge]

Automated review note — AI-generated (Claude), cross-checked by an independent second model against the live diff. Please sanity-check before acting.

Reviewed at head d5e72a7554.

Verdict: REQUEST CHANGES

• BUG — ArduCopter/mode.cpp:570 — Drift mode silently gains simple mode — re-raised, still unaddressed. This was raised on 2026-08-05 at head a5d54f6a50 and has neither been fixed nor answered. apply_simple_mode() is called unconditionally inside Mode::get_pilot_desired_lean_angles_rad(), and mode_drift.cpp routes through that helper. Re-checked at the current head by fetching mode_drift.cpp directly (6393 bytes, so the fetch genuinely succeeded): it still contains zero simple-mode references and still calls the helper, and the PR adds no opt-out. Drift's roll drives a coordinated turn and is inherently body-referenced, so rotating it by the simple-mode heading changes the mode materially. Either exclude Drift or state in the PR description that this is intended. The thread since then discusses the resemblance to Copter: move simple transform into mode class stop using set_control_in #18581 but nobody has engaged with this point.

• ISSUE — ArduCopter/mode.cpp:577 — Implicit application in get_pilot_desired_velocity(). Softened after cross-check: the transform is implicit here too, but this is a design concern rather than a live regression — the callers of this helper already had simple mode. The old design had each mode opt in explicitly, which made the affected set reviewable; the new one makes it implicit.

• NOTE — Tools/autotest/arducopter.py:0 — The new autotest cannot catch the Drift case. It covers AltHold only, so it would not detect the regression above.

Full report: https://uav.tridgell.net/DevCallReviews/2026_08_11/devcall_pr_reviews.html

[IamPete1]

This is surprisingly close to #18581. I think anything we can do to remove uses of the "control in" methods on RC is a good thing.

[rmackay9]

@IamPete1,

That is incredible. It's nearly identical!

[peterbarker]

LGTM

I'm not particularly fussed if there are a few differences in e.g. drift mode and whatnot.

Such a good improvement.