kalibr_allan is a utility repository that provides scripts and tools for calculating IMU noise parameters for use in Kalibr and other IMU filtering systems. While manufacturers typically provide “white noise” values in IMU datasheets, the bias instability and random walk parameters must be determined experimentally. This project enables users to compute those values using Allan variance analysis from recorded IMU data. The workflow involves recording IMU measurements with the device stationary, converting ROS bag files into MATLAB-compatible formats, and then running MATLAB scripts to generate Allan deviation plots. These plots are analyzed to determine noise density and random walk parameters for both gyroscopes and accelerometers. The repository also includes example data and plots from real sensors such as the XSENS MTI-G-700, Tango Yellowstone Tablet, and ASL-ETH VI-Sensor, providing reference points for interpretation.
Features
- Scripts for calculating IMU white noise and random walk parameters
- Workflow for processing ROS bag files into MATLAB-compatible data
- Allan deviation analysis for gyroscope and accelerometer measurements
- Automated fitting of slopes to identify noise density and bias instability
- Includes example datasets and plots for reference sensors
- Supports parallel computation with MATLAB’s Parallel Toolbox
Project Samples
