accelerometer

This is why accelerometers are good friends of gyroscopes: because they can detect excellent "downward" references (at least, when they are in the state of gravity of the earth).

Well, generally speaking. From one moment to another, it receives gravity, linear acceleration, centrifugal force generated by rotation, vibration, noise and, of course, the defects of the sensor.

Therefore, despite the high sampling rate and good precision of MEMS sensors (usually as good as gyroscopes), accelerometer data is still one of the noisy and unreliable data. It has received several "signals" that must be disambiguated. But it is the sensitivity to all these different signals that makes it so universal-it can hear everything.

There are two integration steps of acceleration from the position we really want to know, so we have to summarize a lot of deltas and error lines, from the measured acceleration to the estimated speed, and then we have to do it again to get the estimated position. Errors keep accumulating.

This is why GPS is a friend of the accelerometer: because it will "zero" the increasing position error periodically, just as the accelerometer gently zeroes the azimuth error of the gyroscope.

magnetometer

A good compass should not be ignored. However, like accelerometer, it is usually used as a control method for long-term gyro drift. Knowing which direction is up is more useful than knowing which direction is magnetic north-but with these two, we can know our really good position on the earth.

They are often paired with GPS receivers, because they once again fight against each other's big weaknesses, but they are closely matched in other aspects. One gives a rough excellent position, and the other gives a rough excellent direction.