The relation between several coordinate frames.

[wystephen]

Hi,
The provided dataset contained ground truth, stencil IMU, and iPhone.
How about the transformation between these coordinate frames?
Without such information, It is impossible to reproduce the result of the published paper.

[wystephen]

More specifically, the rotation from phone to reference system [as provided in other comments], represent the rotation from iPhone's body frame to IMU's frame.

python rot_phone2imu = Rotation.from_matrix(np.asarray([[-0.999870, 0.006520, -0.014725], [-0.006843, -0.999734, 0.022022], [-0.014578, 0.022120, 0.999649]])) # from phone to ground truth

However, the real conversion between phones and imu is defined as:
acc_imu = rot_phone2imu * acc_iphone * -1.
gyr_imu = rot_phone2imu * gyr_iphone.

Furthermore, the magnetometer measurements [denoted as iphoneMagX, iphoneMagY, iphoneMagZ ] since like magnetometer measurements represented in reference frame [ global frame of the reference system.].

Could you provide a detail description about how to get the phones real raw output and how to obtain the conversion between iphone's local frame to the global reference frame.