TAP_TU_SyntheticModelGenerator

A tool for generating synthetic Vehicle Routing Problem (VRP) instances by sampling from OpenStreetMap (OSM) data.

Directory Structure

The project contains the following key directories:

• scenarios/: Contains example scenario JSON files that define the input parameters for generating VRP problems
• paper_input_models/: Contains the generated VRP problem files that are created from the scenario files in the scenarios/ directory

Installation

1. Clone this repository:

git clone https://github.com/yourusername/TAP_TU_SyntheticModelGenerator.git
cd TAP_TU_SyntheticModelGenerator

2. Install the required dependencies:

pip install -r requirements.txt

Usage

The osm_synthesiser.py script generates synthetic VRP problems by sampling from OpenStreetMap data. It takes several command-line arguments:

• -s, --scenario_file: Input scenario JSON definition file (required)
• -o, --output: Output JSON file (required)
• -r, --random_seed: Seed for random number generation (optional)
• -c, --cache_directory: OSMnx cache directory (optional, defaults to 'C:\tmp\osmnxcache')
• -h, --help: Print help message

Basic Example

python osm_synthesiser.py -s "scenarios/example_scenario.json" -o "paper_input_models/output_problem.json"

Example with Random Seed

python osm_synthesiser.py -s "scenarios/example_scenario.json" -o "paper_input_models/output_problem.json" -r 42

Example with Custom Cache Directory

python osm_synthesiser.py -s "scenarios/example_scenario.json" -o "paper_input_models/output_problem.json" -r 42 -c "/tmp"

Input Scenario File Format

The input scenario file should be a JSON file with the following structure:

{
    "id_prefix": "example",
    "place_names": ["London, UK"],
    "place_names_choose_one_of": [
        ["London, UK"],
        ["Paris, France"],
        ["Berlin, Germany"]
    ],
    "depot_tags": {
        "amenity": "warehouse"
    },
    "delivery_tags": {
        "shop": "supermarket"
    },
    "pickup_tags": {
        "shop": "convenience"
    },
    "servicing_tags": {
        "amenity": "service"
    },
    "break_locations": {
        "amenity": "restaurant"
    },
    "hub_tags": {
        "amenity": "hub"
    },
    "hub_delivery_tags": {
        "shop": "department_store"
    },
    "hub_pickup_tags": {
        "shop": "mall"
    },
    "target_min_depots": 2,
    "target_max_depots": 5,
    "target_min_vehicles_per_depot": 3,
    "target_max_vehicles_per_depot": 5,
    "target_min_break_at_location": 1,
    "target_max_break_at_location": 3
}

Place Names Configuration

The script supports two ways to specify place names:

1. Single or Multiple Fixed Places:

   • Use the place_names field to specify one or more places
   • Example: "place_names": ["London, UK", "Manchester, UK"]
   • The script will generate a problem that includes locations from all specified places
   • This is useful when you want to create a problem spanning multiple cities or regions

2. Random Selection from Multiple Options:

   • Use the place_names_choose_one_of field to specify multiple sets of places
   • Example:

     "place_names_choose_one_of": [
         ["London, UK"],
         ["Paris, France"],
         ["Berlin, Germany"]
     ]

   • The script will randomly select one set of places from the list
   • This is useful for generating different problem instances in different locations
   • The selection is deterministic when using the same random seed

Example Scenarios

The scenarios/ directory contains several example scenario files that demonstrate different use cases:

1. Basic Urban Delivery Scenario:

{
    "id_prefix": "urban_delivery",
    "place_names": ["New York, USA"],
    "depot_tags": {
        "amenity": "warehouse"
    },
    "delivery_tags": {
        "shop": "supermarket"
    },
    "target_min_depots": 2,
    "target_max_depots": 3,
    "target_min_vehicles_per_depot": 2,
    "target_max_vehicles_per_depot": 4
}

2. Multi-City Pickup and Delivery:

{
    "id_prefix": "multi_city",
    "place_names": ["London, UK", "Manchester, UK"],
    "depot_tags": {
        "amenity": "warehouse"
    },
    "pickup_tags": {
        "shop": "convenience"
    },
    "delivery_tags": {
        "shop": "supermarket"
    },
    "target_min_depots": 3,
    "target_max_depots": 5,
    "target_min_vehicles_per_depot": 3,
    "target_max_vehicles_per_depot": 5
}

3. Random City Selection:

{
    "id_prefix": "random_city",
    "place_names_choose_one_of": [
        ["London, UK"],
        ["Paris, France"],
        ["Berlin, Germany"]
    ],
    "depot_tags": {
        "amenity": "warehouse"
    },
    "delivery_tags": {
        "shop": "supermarket"
    },
    "target_min_depots": 2,
    "target_max_depots": 3,
    "target_min_vehicles_per_depot": 2,
    "target_max_vehicles_per_depot": 4
}

Output Format

The script generates a JSON file containing the synthetic VRP problem with the following structure:

{
    "data": {
        "jobs": [...],
        "vehicles": [...],
        "hubs": [...]
    },
    "_id": "example_42"
}

The generated files are stored in the paper_input_models/ directory, with filenames based on the scenario's id_prefix and the random seed used.

Notes

• The script uses OSMnx to fetch data from OpenStreetMap, so an internet connection is required.
• The cache directory is used to store downloaded OSM data to avoid repeated downloads.
• The random seed can be used to reproduce the same problem instance.
• When using place_names_choose_one_of, the script will print the selected place names during execution.
• Multiple place names in place_names will create a problem that spans across all specified locations.
• The script will fetch data for each place name separately and combine them into a single problem instance.
• Example scenarios in the scenarios/ directory can be used as templates for creating new scenarios.
• Generated problem files in paper_input_models/ can be used as input for VRP solvers.