Rover: add arm-once-after-prearms-pass functionality

[peterbarker]

Add option to Rover to arm the vehicle once, once prearm (and arm!) checks have passed.

This is intended to be an alternative to having an ARMING_REQUIRE of 0, which can cause all sorts of fun which is best avoided.

Also adds an "REQUIRE_POSITION" as a DriveOptions setting for Rover, the same as Copter's "REQUIRE_POSITION" functionality.

Tested in SITL.

autotests coming, assuming this doesn't get insta-killed...

[IamPete1]

Doesn't have the advantage of your get_last_prearm_checks_result but this does work:

-- Arm the vehicle one time once arming checks have passed
local function update()

    if arming:is_armed() then
        -- Kill script once armed
        return
    end

    if arming:pre_arm_checks() and arming:arm() then
        -- Kill the script once it works
        return
    end

    -- try again in 5 seconds
    return update, 5000
end

return update, 5000

[rmackay9]

I'd like to avoid this new parameter if possible. I think the user could achieve essentially the same thing by setting the INITIAL_MODE parameter to a mode that requires a position estimate?

[peterbarker]

The point here is to require that the vehicle has position in all modes. So the user can't arm the vehicle at all until it knows how to get home.

[rmackay9]

yes, I definitely understand the need for this now and we have the same option in Copter. I just wonder why we put it into FLIGHT_OPTIONS instead of ARMING_OPTIONS as part of PR #27367

[peterbarker]

yes, I definitely understand the need for this now and we have the same option in Copter. I just wonder why we put it into FLIGHT_OPTIONS instead of ARMING_OPTIONS as part of PR #27367

I think we did discuss at DevCall - but I can't find any notes on it.

It probably should be in arming options. Should I move it? This will require some awkward code in Copter if we want people to start using the bit in ARMING_OPTIONS. But I think we should.

[rmackay9]

Hi @peterbarker yes, I'm sorry but I agree that we should move it. It's new for 4.6 so I think we could get away without the parameter conversion if we backported it before the official release

[peterbarker]

.... actually, I think the problems revolved around Plane.

If we ever allow Plane to arm without position (@Ryanf55 ...) we would have issues around retaining the default for Plane. parameter defaults and bitmasks are a mess - if someone already has set ARMING_OPTION in their parameter storage then you can't have sensible default values on a per-vehicle basis (or indeed any non-zero bit default).

[rmackay9]

For copter and rover the default is zero (e.g. require GPS for arming is false) but I guess plane would like that to default to 1. It's sad to put the option in the wrong place because of this though. I understand the issue but.. it's hard to accept doing the wrong thing for a small technical reason

[Ryanf55]

Good details, I need time to digest, but am swamped at work. If you aren't in a rush, I can discuss in 2 days.

[peterbarker]

PR to move the bit into AP_Arming: #29076

[rmackay9]

Thanks for this, it would be great if we could eventually remove the arming-required =0 option