Gluonpilot

I'm sorry Mezuzu, in this case I didn't understand your question. Please try to be more clear

Hi Tom,

My question is about the comment you made on how to correct the Gyrobias.

You say that there are two ways

1. Change Gyrobias watching the IntegratedRollAngle - AccRollAngle
2. Change compared with AccRollAngle IntegratedRollAngle

My question is how they relate IntegratedRollAngle - AccRollAngle that is in (rad / s or ° / s) with Gyrobias (volts).

To locate you but I tell you that I'm working with IDG-500 gyroscope and accelerometer ADXL 325, I am integrating with the Kalman filter. As I said in the previous message, everything works fine if the data are few (1000), the estimation of the angles roll, pitch and yaw works very well because, I estimate the bias of the sensors to the average of the first 100 data taken with no motion sensors.

The problems begin with the time that the data sample increases and the bias of the sensors changes. So I want to know how to estimate the gyrobias.

Regards

mezuzu

IntegratedRollAngle, AccRollAngle are in degrees or radians

the gyro bias can be in volt, unitless, deg/s or rad/s depending on your implementation

raw_gyro = raw_gyro - raw_bias
or after scaling
gyro_rad_s = gyro_rad_s - bias_rad_s
or
...

It just depends on your implementation!

Hi Tom,

Thanks for your answers.

IntegratedRollAngle and AccRollIntegrated are in ° / sec.

The IDG-500 gyroscope delivery Gyro values in Volts, Gyrobias therefore must be in volts, so that they can subtract.

The sensitivity is in mV / ° / s. Replacing in the equation

IntegratedRollAngle = IntegratedRollAngle + (Gyro - Gyrobias) / Sensitivity

all factors are in ° / sec.

Right now I'm manually editing the bits of Gyrobias met without motion sensors. With this change IntegratedRollAngle values positively but it is clear that this method is not the most optimal.


Regards

Mezuzu,

IntegratedRollAngle should be in degrees (as the word "angle") says.
Gyro = rotation rate = °/sec

Hi Tom,

You're right I was wrong to write it .

Regards

Hi Tom,

I just want to know a method to find the angle matches the actual rotation. This depends entirely on the GyroBias.

I believe the method I'm using is not the most optimal, since I'm changing the bits manually adjust and hopefully get the angle. This method I can not implementers in matlab, because the bits changed in each experiment.

Thanks

Hi Mezuzu, you need to calculate the difference between the integrated angle, and the one calculated from the accelerometers. If the integrated one is smaller, adjust the gyro bias small bits at a time...

Hi Tom,

I made the comparison between the angles obtained with accelerometers and those obtained with the gyros. The difference is quite large, with an accelerometer get 30 degrees (real angle) and the gyroscope -150 °. Modify gyrobias bits, changing bit more influential, but all I get is to decrease the angle obtained from the gyroscope to -100 °.
The modification of the bit carried generating random numbers in an interval containing the initial gyrobias, met with the gyroscope free motion.

I wanted to ask if you only had the gyros IDG-500, which would be the way to get the bias for each gyro, for different angles roll, pitch and yaw.

Regards

Mezuzu

Hi Tom,

The previous question, I do because in the kalman filter I'm doing, the estimation of the angles is with the gyros, and I do measurements with accelerometers.

Given the circumstances in which the accelerometers give erroneous readings (gravity affects them more than 1g) the gyroscopes must be able to continue estimating the angles roll, pitch and yaw.

Thanks

Mezuzu