Usage

Translating tachograph data into HTN knowledge

We provide Python scripts to transform a CSV tachograph into HTN domains and problems:

python ./src/parsers/fromCSVtoPLAN.py <csv_file> <plan_output_folder_path>

and then, given the <driver_name> for the driver sequence to tag:

python ./src/parsers/fromPLANtoPDDL.py <plan_output_folder_path>/event-log-<driver_name>.plan <problem_output_folder_path>

Please see our data examples as input files require specific formatting.

Tagging Driver Activities

With a previously defined problem, just call the SIADEX planner with the following command:

./planner/planner -d hpdl/domain.pddl -p <problem.pddl> -o <output_path>

or calling the script:

./scripts/runPlanner.sh <domain.pddl> <problem.pddl> <ouput_path>

A tagged log in TSV format will be outputted.

Suggesting driver activities

The HPDL domain is also capable of suggesting driver activities based on Zeno-Travel while complying with the HOS regulation. After trying to recognize an optional driver log, new Drive/Break/Load/Unload/Refuel activities will be included at the end indicating how to deliver the specified packages to their destinations.

Please see some of our examples (like this one) for how to define this problems.

Clustering

We provide a K-Means model pre-trained using Doc2Vec (also called Paragraph Vector model), which can be loaded from the src/model folder. Each of the 25 centroids can be found in human readable format in the out folder.