AP_DDS: Provide airspeed for planes

[tizianofiorenzani]

Airspeed for planes

The DDS interface does not provide the airspeed, which is an essential information for planes (subs?).

Solution

• Publish the airspeed using the same /ap/twist/filtered topic, with a dedicated frame_id = "airspeed".
• Leave the frame_id empty for standard Inertial/Body

Alternative

• Publishing a TwistStamped message for just one element seems like a waste of bandwidth.
• We could create a dedicated /ap/airspeed/ topic of type std_msgs/Float32, as the timestamp might not be crucial.

Platform
[ ] All
[ ] AntennaTracker
[ ] Copter
[ x ] Plane
[ ] Rover
[ ? ] Submarine

[Ryanf55]

Airspeed would be great to add to the interface. Airspeed doesn't have a rotational component though, so twist seems overkill and wasteful of data. Did you consider publishing just a geometry_msgs/msg/Vector3 with the body frame component of estimated airspeed? Plane's zero sidesplip assumption means it would only be populating the X component, but copter could populate more fields.

[tizianofiorenzani]

I like this idea, we could use the three components to publish.

• x: airspeed
• y: angle of attack (positive when flow comes from underneath)
• z: sideslip angle (positive when flow comes from left)
• topic name could be /ap/airspeed, where the frame_id = wind_axis

[Ryanf55]

Sideslip is only for plane. Why not just XYZ relative to body frame.

[tizianofiorenzani]

You mean airspeed in body frame? That would work. length of vector is what we call "airspeed", which most of the times will result in only the x component unless you have AOA and sideslip measurements, or they are estimated internally.

[Ryanf55]

You mean airspeed in body frame? That would work. length of vector is what we call "airspeed", which most of the times will result in only the x component unless you have AOA and sideslip measurements, or they are estimated internally.

Yea, that's what I meant.

[tizianofiorenzani]

I can work this out in a PR and we can move ahead from there.

[tizianofiorenzani]

Well, from inspecting the code it seems pretty easy to obtain the TAS vector:

    Vector3f airspeed_vec_bf;
    if (AP::ahrs().airspeed_vector_true(airspeed_vec_bf)) {
        // we are running the EKF3 wind estimation code which can give
        // us an airspeed estimate
        return airspeed_vec_bf.length();
    }

[Ryanf55]

Awesome, at what rate do you think it should be published?

[tizianofiorenzani]

Here is the PR: #28319

[tizianofiorenzani]

Awesome, at what rate do you think it should be published?

I publish it within the same velocity timer, if airspeed is available.