I am aware that most launch vehicle companies use strapdown IMU systems, consisting of gyros and accelerometers. I was thinking if one could use a magnetometer and accelerometer to get data (like they do in mobile phones), since calibrating a gyroscope seems like a difficult task.
As a disclaimer: I do not have specific information nor any connection to those companies, I would expect such details not to be public and trust no one to be a reliable source for industry intellectual property except the specific person who developed and deployed the technology.
There are plenty references and methods for calibrating an magnetometer as well as calibrating IMUs, I do remember doing both these tasks.
The big bottomline I've got from my limited practical experience was:
The magnetic field in any place is subject to a lot of disturbances. This gets worse around electronic equipment and metal artifacts. So even if you had a perfect mangnetometer which required absolutely no calibration, for any purpose that would require having a comparison reference (such as attitude determination), the achievable accuracy would be very limited. Meanwhile, good inertial systems achieve fairly good performance
As magnetometers do require calibration, that should preferably be performed when installed on the target platform, the platform itself is also required to move. And as you can't play much with a rocket's position, (same as you can't do so with an airplane) then maybe you shouldn't trust magnetometers.
Given the earlier points, I am aware of some aircraft which carried magnetic based attitude determination systems as independent backup for the IMU, the reasoning being that someone someday required an independent fallback method of attitude determination (though allowing for poor attitude estimation) for safety reasons. But notice that for a small aircraft the pilot is also available, but for a rocket that needs to guide itself its instruments should be even more reliable.
I am aware of some rocket designs which performed stationary alignement (I've worked with in-flight alignment but that was more pure research). This procedure is well known and does allow calibrating a gyro. It's complex and delicate indeed. But as far as I was informed, this is both a necessary and there are well-mastered methods for this application.
Hence, even if calibrating gyros is difficult, it is also mandatory. While calibrating magnetometers is not only difficult but it's ineffective and unnecessary.