A collection of work regarding calibrating multiple IMUs using Kalman filter frameworks This package simulates multiple IMUs in a calibration EKF framework that produces body acceleration and angular rate errors that outperform a single IMU or a complementary filter using multiple IMUs.
By randomly assigning calibration errors, this simulation can run a Monte-Carlo simulation to show the error distribution and overall stability of the filter.
Running a single instance of the filter is very easy, using the run_sim()
function. A bump time can also be input to randomly change the extrinsic
calibration of the IMUs to simulate a jerk or bump of one of the sensor mounts
An example of running a Monte-Carlo simulation can be found in run_sim.ipynb
which utilizes a multiprocessing pool to mapping the simulation outputs to get
the EKF and CF results.
Plotting and statistical results can be generated using the
plot_*() and evaluate_ekfs() functions respectively.