readme_main

---------------------------------------------------------------------------------------
General:
Title of dataset: "Experimental-based Cramér-Rao Lower Bound Estimation of Triaxial Accelerometer Calibration Methods using Monte Carlo Simulation".

Date: 18/07/2025

Location: Centre for Defence Engineering and Physical Sciences, Cranfield University, Defence Academy of the United Kingdom, Shrivenham, Swindon, SN6 8LA, United Kingdom

Principal investigator: Dariusz Maton. Researcher at Centre for Defence Engineering and Physical Sciences, Cranfield University. Supervised by Prof. John Economou, Dr. David Galvao Wall, Dr. Irfan Ullah Khan, Robert Cooper, Simon Trythall and Stuart Kitching.

Funding source: This work was supported by the EPSRC iCASE grant reference EP/S513623/1, BAE Systems and Cranfield University.

---------------------------------------------------------------------------------------
Method:
Simulation:
Monte Carlo simulations were run in MATLAB to assess the calibration accuracy of two static triaxial accelerometer calibration methods, the six-position and multi-position methods (SPM and MPM respectively). MATLAB code for running the simulations is included in this data. Note it requires MATLAB's Navigation Toolbox to access some of the functions.
Experimental:
Experimental data was gathered for four LSM6DS3 inertial measurement units mounted on a two-axis gimbal sampling 26 positions (including the cardinal positions). The .CSV files containing the experimental data had a total duration of one hour with ambient temperature change less than 0.3°C and relative humidity change less than 0.5% over this time. Quality of this data was assured by averaging measurements at each position to reduce the impact of noise in the measurements as well as adding delays to ensure the sensors were only sampled in the steady state on the gimbal. The average of 100 samples was also taken to replicate the conditions in the simulations.

---------------------------------------------------------------------------------------
Licenses:
CC BY 4.0

---------------------------------------------------------------------------------------
File name structure and file format:
MATLAB script files are:
MonteCarloMultiPosition.m and MonteCarloSixPosition.m
All other .m files are supporting MATLAB functions.

To run Monte Carlo simulations the main MATLAB scripts are:
MonteCarloSixPosition.m and MonteCarloMultiPosition.m

All other .m files are supporting functions relating to running these two scripts.

Folder "CRLB" contains supporting .m functions relating to the derivation of the Cramer-Rao lower bounds for the methods.

Folder "kinematic_trajectories" contains kinematic data required to simulate the SPM and MPM in MATLAB. These are given as time series data in .CSV format. They give the kinematic variables expressed in the navigation reference frame. The sub-folder "mpm_different_positions" contains kinematic data relating to the Monte Carlo simulation using different types of positions in the MPM and different order. The diagram in that folder shows each of the four cases investigated for ease of understanding.

Folder "multi_position" contains .m functions required for implementing the MPMs. A separate .txt file gives links to third party implementations of the methods of Frosio et al., Merayo's method and MVEE. The folder also includes the experimental and simulation data relating to the study of their sensitivity to parameter initialisation.

Folder "six_position" contains .m functions required for implementing the SPMs.

All sources code is provided in .txt format but can be copy and pasted in MATLAB and saved as .m files.
Kinematic trajectory data and any experimental/simulation results are given as .CSV files.
Diagrams are given as .PNG files.

--------------------------------------------------------------------------------------
Data-specific information:
For the kinematic trajectories, .CSVs have the header:
acc_x (m/s2): acc_y (m/s2): acc_z (m/s2): gyr_x (rad/s): gyr_y (rad/s): gyr_z (rad/s): t(s):

Note: acc and gyr are acceleration and angular velocity of sensor relative to the navigation frame, expressed in the navigation frame.