Unit mismatch in Dynamic Bias PSD calculation units are not matched with equation.

[venkat057]

Hi,

In conversions/imu_si_errors.m at line 95 and 101 while calculating Dynamic Bias PSD calculation units are not matched with equation.
imu_si.ab_psd = imu_si.ab_dyn ./ sqrt(imu.ab_corr); % m/s^2/root(Hz)
--- Actual units will be % m/s^2/root(time) ....
--- correct Equation will be imu_si.ab_psd = imu_si.ab_dyn .* sqrt(imu.ab_corr); % m/s^2/root(Hz)
Please check .
same for gyro bias PSD.

In kf_update.m why below equation is being used instead of Joseph stabilized version which is correct as per many references. Is there any advantage with it.?

% Step 5, update the a posteriori covariance matrix, Pp
kf.Pp = kf.Pi - kf.K * kf.S * kf.K'; % Eq. 3.10 from Särkkä

[rodralez]

Hello,

The only problem is that in fact the PSD unit is √PSD (root-PSD), not just PSD.

According to Groves, p. 156, root-PSD is:

1 mg/√Hz = 9.80665 * 10^-6 m * s^-1.5 
1 °/hr/√Hz = 4.848 * 10^-6 rad * s^-0.5

Where,

syms Hz s m

Hz = 1/s;
eq1 = m / s^2 / (Hz)^0.5;
eq2 = simplify(eq1)
eq2 =
 
m*(1/s)^(3/2)
 

So,

m / s^2 / (Hz)^0.5 == m*(1/s)^(3/2) == m*s^-1.5

Same results can be achieved following the previous steps for 1 °/√hr