Calibration Process
Analytical techniques used to determine the fixed or slowly varying offsets in a measurement device ensure that the raw data can be corrected for systematic errors. Successful sensor bias parameter identification allows for the removal of the zero-point error from the signal. This step is required before a sensor can be used in a high-precision navigation system.
Static Testing
The process often involves placing the sensor in a series of known orientations relative to gravity. For sensor bias parameter identification, a multi-position tumble test is used to isolate the bias from the scale factor. The results are averaged over a long period to reduce the impact of random noise.
Mathematical Estimation
In many systems, this identification happens in real time using a kalman filter. The sensor bias parameter identification is performed by adding the bias as a state in the estimation vector. As the system moves, the filter compares the sensor output to other references to solve for the unknown offset.
This requires the system to undergo specific motions that make the bias observable. For instance, a gyroscope bias is best identified when the vehicle is stationary. Once the parameters are identified, they are used to compensate the raw data.
Long Term Stability
The identified parameters often drift due to temperature changes or aging. Frequent sensor bias parameter identification is necessary to maintain the accuracy of the overall system over its operational life.
