This repository contains the code for our paper:
Multimodal Asynchronous Kalman Filter for Monitoring Unstable Rock Slopes
Lukas Schild, Thomas Scheiber, Paula Snook, Reza Arghandeh, Stig Frode Samnøy, Alexander Maschler, et al.
Published in Geomatics, Natural Hazards and Risk, Volume 14, Issue 1, 2023
DOI: 10.1080/19475705.2023.2272575
This paper proposes a Multimodal Asynchronous Kalman Filter (MAKF) for fusing heterogeneous sensor data to monitor unstable rock slopes. The method uses a two-stage architecture: multiple local Kalman Filters (one per sensor) compute estimates at each sensor's native sampling rate (asynchronously to allow for mssing/corrupted measurements), which are then combined by a global Kalman Filter that synchronizes the outputs.
The data directory contains all data used in the paper and is necessary to run the demo file.
MAKF_demo.ipynb contains a detailed explanation of the data-preprocessing and the application of the filters.
asynchronous_kalman_filter.jl contains the implementation of a Kalman Filter processing the measurements for the different sources (sensors).
The files extensometer_pre.jl and totalstation_utils.jl contain pre-processing code related to the Extensometer and Total Station data repsectively.