Barometer

[dkokron]

Firstly, thank you for taking the time for publishing this work. It has been very helpful to me.

I have been playing with the examples/synthetic-data/navego_example_synth.m case. I've added two 15 second GNSS outages at 145.75s (straight segment) and 306.0s (just before and during a turn) into the simulation.

I've also added the ability to use the integrated barometer found on the ADIS16488 IMU to update estimated height (h_e) during the entire simulation. I followed http://users.cecs.anu.edu.au/~Jonghyuk.Kim/pdf/KimPhDWeb.pdf (Section 5.5.1) for this addition.

My implementation does a good job of keeping the height close to the reference during the GNSS outages, however there is a large spike in height when the GNSS data returns. Blue marks the simulation of the ADIS16488 IMU without barometer aiding, red depicts the same IMU with barometer aiding and the magenta dots mark availability of GNSS data.

As far as I can tell, the Kalman filter is drawing close to the GNSS data upon its return even though its Pi should be rather large (being multiplied by 15.4 seconds instead of .2).

Could this result from not updating the vertical velocity during the barometer cycles?

Also, there is significant horizontal track differences between the two simulations.

Could this result from not resolving the barometer height measurement (in NED) into the body frame?