Skip to content

README.md

Latest commit

 

History

History
77 lines (54 loc) · 2 KB

README.md

File metadata and controls

77 lines (54 loc) · 2 KB

Noise Generation Example

This example creates a sequence of samples for an accelerometer and gyroscope based on the https://www.analog.com/media/en/technical-documentation/data-sheets/adis16448.pdf, a commonly-used MEMS IMU.

The sequence of samples can then be evaluated with Allan Variance to determine that the noise generation in ign-sensors is working as expected.

To evaluate noise, first build and run.

mkdir build
cd build
cmake ..

Ubuntu Or MacOS

make

This will generate the sensor_noise executable under build.

Windows

cmake --build . --config Release

This will generate the sensor_noise.exe executable under build\Release.

# Ubuntu or MacOS
./sensor_noise

# Windows
.\Release\sensor_noise.exe

This will generate two series of samples, one for accelerometer and another for gyroscope.

To use the analysis script, first install Allan Tools:

pip install --user allantools scipy

Then execute the script:

# Ubuntu o MacOS
../plot_samples.py

# Windows
python.exe ..\plot_samples.py

This will produce two graphs: the Allan Deviation plots for both the simulated accelerometer and gyroscope. The values on these graphs should correspond closely to the inputs:

  • Gyroscope Allan Plot:
    • sigma_N: Gyroscope Noise Density
    • sigma_K: Gyroscope Random Walk
  • Accelerometer Allan Plot:
    • sigma_N: Accelerometer Noise Density
    • sigma_K: Accelerometer Random Walk

While this technique is used here to validate the operation of the algorithm, it could also be used to estimate the noise parameters for a real IMU.

References: