Split-KalmanNet

Split-KalmanNet: A Robust Model-Based Deep Learning Approach for State Estimation

Simulation parameters are in 'config.ini' file.

https://arxiv.org/abs/2210.09636

Run experiments

  • IMPORTANT:

    • Check that you have the correct parameter settings in config.ini
    • Toggle TRAIN = True or False at the beginning of the main scripts

Uniform Circular Motion

Convergence

python \(SyntheticNL\)\ main.py
python \(SyntheticNL\)\ plot_loss.py

Noise heterogeneity and non-linearity

python \(SyntheticNL\)\ main_varied_noise_heterogeneity.py
python \(SyntheticNL\)\
plot_varied_noise_heterogeneity.py

NCLT Dataset

python \(NCLT\)\ load_data.py
python \(NCLT\)\ main.py
python \(NCLT\)\ plot_trajectory.py