Magnetometer: correct field calculation

[srmainwaring]

🦟 Bug fix

Summary

Correct the magnitude and direction of the magnetic field calculation in the Magnetometer sensor.

Details

• The magnetic intensity table has units centi gauss not centi tesla.
• Add a conversion to publish the magnetic field in tesla.
• The formula to calculate the field from the declination, inclination and intensity is in the NED frame.
• Convert the field orientation to the ENU frame to conform to the Gazebo world frame convention.

Context

The issue was found during the development of external sensor support for ArduPilot using DDS to subscribe to ROS based sensor topics (in this case magnetic field strength published from Gazebo via the ros_gz bridge).

ArduPilot SITL and Gazebo use similar calculations to simulate the earth's magnetic field at a given location (lat, lon). There was an inconsistency between the two calculations which can be attributed to a combination of incorrect units and frame convention.

The attached notebook reproduces both calculations in Python for comparison.

earth_mag_field.ipynb.zip

Checklist

• Signed all commits for DCO
• Added tests
• Updated documentation (as needed)
• Updated migration guide (as needed)
• Consider updating Python bindings (if the library has them)
• codecheck passed (See contributing)
• All tests passed (See test coverage)
• While waiting for a review on your PR, please help review another open pull request to support the maintainers

Note to maintainers: Remember to use Squash-Merge and edit the commit message to match the pull request summary while retaining Signed-off-by messages.

[azeey]

There's a stalled PR #2336 that was fixing the units. There we decided to add a flag to enable this so as to not break behavior. Can we do the same here?

[srmainwaring]

There's a stalled PR #2336 that was fixing the units. There we decided to add a flag to enable this so as to not break behavior. Can we do the same here?

Thanks for the link to the PR, missed that.

The problem with the approach of continuing with the wrong units and orientation by default is that it creates issues downstream - particularly once the data is pushed into the ROS data space where the assumption is that it will be standards conforming.

Perhaps we could do the following:

Add params, which for garden and harmonic default to current behaviour:

<use_units_gauss>true</use_units_gauss>
<use_earth_frame_ned>true</use_earth_frame_ned>

and mark this behaviour as deprecated. Then switch to standard conventions as default if omitted (i.e. both false) in a future release.

Updated in: e8f1628

[azeey]

Adding explicit parameters for this sounds good to me. As far as changing the default, I see that https://www.ros.org/reps/rep-0103.html NED for outdoor systems. Shouldn't NED be the default then?

I just have one minor comment. Thanks!

[srmainwaring]

Shouldn't NED be the default then?

The reasoning is that for Gazebo the world frame is ENU, so to be consistent with the orientation given by other sensors (say NavSat when spherical coordinates are provided) I think we need to use ENU when setting the mag field in the world frame.

I'll need to review REP 103 again for the ROS msgs, if the convention is NED rather than ENU that should probably be handled in the ros_gz bridge.

Update

REP 103 suggests

For short-range Cartesian representations of geographic locations, use the east north up [5] (ENU) convention:

X east
Y north
Z up

If magnetic field data is regarded as geographic data the default frame (no suffix) is ENU. To use the NED frame (which is what aerial vehicles use in flight controllers) the ROS frame_id would add the _ned suffix.

[azeey]

Okay, thanks for the explanation.