[continue:] Moving towards multi-sensor-per-same-measurement fusion

[kamilritz]

This PR is exactly the same as: #652,
but added as a separate branch to ECL. Such that we can test the PR's compatibility with the Firmware's CI.

[priseborough]

Approved, but please delay merging until corresponding upstream PX4/Firmware changes PX4/PX4-Autopilot#13127 are approved and ready to merge so that merging can be coordinated.

[mhkabir]

Question: why do we even use the std. deviation in the estimator interface?

1. PX4 side uses variances (from cov matx)
2. Fusion uses variances

So we just end up uselessly square rooting and then squaring.

[mhkabir]

Suggested change

	}else{
	} else {

[kamilritz]

That is only the tip of the iceberg. At some point we really need a formatter.

[kamilritz]

@mhkabir I also noticed that and want get rid of the std deviation. I am in the process of cleaning this and some other things up. But these changes are based on top of this refactor. I am eager to get them in after this PR.

[mhkabir]

@kamilritz What is missing here? Let's bring this in soon.

[kamilritz]

@mhkabir Glad you feel the same. Waiting on the firmware PR to be ready.

[xdwgood]

@kamilritz https://github.com/PX4/ecl/blob/master/EKF/vel_pos_fusion.cpp#L168

This is just KP, why does the code say it is KHP, where does H come from? thank you very much 😃

Is it because the H matrix is 1?

[kamilritz]

Is it because the H matrix is 1?

Yes. To be a bit more precise. It is a vector of size 1x24 full of zeros with a single 1 at the location of the measurement.