Collaboration between ESSEC and Google OR about middle-mile logistics. This repository is supposed to contain only source code.

Instance Generator for Middle Mile Logistics Optimization

This repository provides a tool for generating diverse instances of logistics networks and parcels, aiming to facilitate research and development in the field of middle mile optimization.

Objective

This project seeks to create a rich dataset of instances (LogisticNetwork, Parcels) that can be utilized for:

Benchmarking Optimization Approaches:

Evaluating and comparing different methods for solving the middle mile optimization problem.

Training Machine Learning Models:

Training models to predict and optimize middle mile logistics operations.

Design Overview

The generated instances will represent a variety of real-world scenarios with diverse features like size and structure.

Generation Details

• Hub Graph: A directed acyclic graph (DAG) representing hubs is generated using the Barabási-Albert model, mimicking real-world network properties.
• Lines: Lines are constructed as contiguous paths within the hub graph, choosing random neighbors at each step with a length restriction.
• Rotations: Each line has a randomly assigned duration between connected hubs.

Using the library

Build

$ cmake -S. -Bbuild -DCMAKE_BUILD_TYPE=Release
$ cmake --build build -j 8 -v

Run

$ ./main instance_test -hubs_number 100 -shipments_number 20 -nbr_lines 10 -max_length_line 10 -max_numb_rotations_per_line 10