Correcting imu.heading value through imu.settings()

[deopark]

When adjusting imu.heading values through imu.settings(),
is it more effective to use acceleration_correction and acceleration_threshold in addition to headings_correction and angular_velocity_threshold?

I'm asking this because I'm not familiar with how the hub internally computes heading.

Should I just focus on angular_velocity_threshold and headings_correction?

Thanke you.

[Debenben]

The imu measures 6 values: Acceleration in x,y and z direction and angular acceleration around x,y and z axis. Acceleration includes gravitational acceleration and is used to obtain roll and pitch angles. Angular acceleration is used to get the change in heading.

• acceleration_correction improves roll and pitch, it does not improve your heading.
• angular_velocity_scale improves heading in x,y and z direction. headings_correction is the same in one dimension if your hub has a given roll and pitch angle which does not change.

The imu calculates the heading $h$ as a function of time $t$ from the angular acceleration $a$ using
$$\quad\displaystyle h(t) = \iint a(t) dt dt.$$
For an initial heading $h_0$, an initial angular velocity $v_0$ and a constant angular acceleration $a(t)=a_0$ you get
$$\quad\displaystyle h(t)=\frac{a_0}{2} t^2 + v_0 t + h_0$$.
If you do nothing the measured heading gets worse and worse because it accumulates errors which grow in time. To avoid this, it recalibrates by setting $v_0$ to zero whenever stationary() == true.

• angular_velocity_threshold and acceleration_threshold can be increased or decreased depending on your usecase to make sure it calibrates as often as possible but only if the hub is indeed stationary.

[Debenben]

PS: Acceleration is sensitive to vibrations like a truck driving by, someone walking nearby or some other motor next to the hub. To get the best possible value for acceleration_threshold in your environment, you could start with a low value and make a program that leaves the hub stationary and keeps increasing the threshold whenever stationary()==false.