AdaptiveHGPMPC_ECC25

Instructions for setup:

• Clone the repository to your local machine.
• Install the requirements with pip install -r requirements.txt
• Run the setup file with pip install -e . in the project root directory.
• Update consts/vars.py's data_dir variable with the path to the cloned data_dir directory as this is necessary for some experiments.
• You should now be able to run the jupyter notebooks. In case of any issues, print sys.path in the notebook to ensure the src directory can be found.

Notebooks:

ECC25_adapmap_results:

This notebook visualizes the results from the adaptive mapping section of the paper.

The first section demonstrates our proposed adaptive likelihood-prior trade-off scheme on a toy example.

The rest of the notebook then demonstrates how the proposed approach also transfers well to the closed-loop control task along with an ablation example of the adaptation speed on control performance.