Understanding Sensor Bias (offset)


When looking at the inertial sensor data from the gyroscopes and accelerometers you can see that there is often a small offset in the average signal output, even when there is no movement. This is what is also known as Sensor Bias.

Figure 1 - MT Manager screenshot of the inertial data angular velocity all axis


Figure 2 - Sensor bias of the Z axis angular velocity


Why is there Sensor Bias?

The gyroscopes and accelerometers used in the MTi's are MEMS (Micro Electro Mechanical Systems) sensors. The physical properties of these sensors change over time which results in different values over time. Depending on sensor usage and time the internal sensor biases will increase.

What about the calibration of the MTi's that Xsens performs?

The physical property of each sensor is different and to correct for each sensor difference the MTi's are calibrated. If no calibration would be performed the biases would be much higher. Even though the MTi's are calibrated and that the MTi uses advanced algorithms to correct for the bias, the physical properties of the sensors change which results in a sensors bias.

 How does the sensor bias affect my measurement?

There are two use cases:

  1. Outputting inertial data
  2. Outputting orientation

Outputting inertial data

When outputting the inertial data (gyroscope and accelerometer data), in most cases you will see sensor bias in your output. You can measure the biases when the MTi is laying still. Remember that the biases change, so there is no fixed value you can use to correct the bias for a longer period of time. 

For more information please see the Gyroscope and Accelerometer specifications.

Outputting orientation

If not corrected and you would use only the gyroscopes to calculate the orientation the orientation would drift based on the sensor bias. For example; using the sensor bias of 0.1 deg/s as shown in figure 2 it would mean that the orientation would drift 0.1 deg/s. See figure 3 below.

Figure 3 - Orientation drift shown in MT Manager without bias correction of 0.1 deg/s over 60 seconds

How does the MTi correct for sensor bias in orientation output?

The MTi's use sensor fusion to correct for the sensor biases in the orientation. The MTi uses Gyroscopes, Accelerometers and the Magnetometer (also GNSS for MTi-G-710). Combining these sensors gives the MTi the possibility to detect and correct for sensor biases. It does this e.g. using accelerations, the gravitational pull of the earth, angular velocities and the magnetic field.

For more information see section Xsens Kalman Filter (XKF3i) in the MTi User Manual.


Using the magnetic field to estimate gyro bias

There are two situations:

  1. Stable magnetic field
  2. Unstable Magnetic Field

Stable Magnetic Field

The magnetometer plays an important role in correcting the sensor biases. In the case the gravitational pull cannot be measured in an axis (like shown in figure 3) the magnetic field can be used to correct for the sensor bias. A filter profile which uses the magnetometer needs to be selected and you would need to perform a Magnetic Field Mapping (magnetic calibration).

Important is that when using the magnetometer for correcting the orientation is that the magnetic field must be reliable enough. A non homogeneous magnetic field (disturbed) the magnetic field cannot be trusted, using a non homogeneous magnetic field will result in an incorrect orientation. The MTi will correct the orientation based on the magnetic field even if it is incorrect.

For more information see section Using the earth magnetic field to stabilize yaw in the MTi User Manual.

Unstable Magnetic field

When the magnetic field cannot be used to correct the orientation there are 3 options how to correct for sensor biases:

  1. Gyro bias estimation (No Rotation)
  2. Select filter profile not using the magnetometer
  3. Enabling Active Heading Stabilization (AHS)

For more information see section Estimating the gyro bias in magnetic disturbed environment in the MTi User Manual.

Was this article helpful?
0 out of 0 found this helpful
Do you have a question? Please post your question in our Community Forum