Vehicle Routing Problem

This repository contains a framework for algorithms to solve the Vehicle Routing Problem (VRP) approximately via Local Search Heuristics. It is a wrapper for the following four algorithms implemented by emarkou in this repository:

• Nearest Neighbor
• Local Search with intra- relocation moves
• Local Search with intra- and inter- relocation moves
• Tabu Search

The framework runs all algorithms on the given problem instance(s) and returns the best solution.

Features

The framework additionally offers

• A simple simulation environment management system
• An InstanceHandler to provide functions to systematically generate Problem instances
• A plotter for visualizing the solutions

Usage

Make sure the EXPERIMENTS_HOME environment variable is set and points to a directory that will then contain the simulation results and plots.

To build, make sure you have Maven installed and run the following command in the root directory

mvn package

After the installation you can run it on a randomly generated instance with

java -cp target/vrp-1.jar com.vrp.app.VRP <number vehicles> <number locations> [nosave]

To bypass the storing of the input and solution files the third, optional parameter can be set to nosave. An example call thus looks like this

java -cp target/vrp-1.jar com.vrp.app.VRP 3 70 nosave

Plotter

The plotter will automatically generate the plots for all instances in the input/ directory if a corresponding solution file exists in the output/ directory. Run the plotter with

 python3 plotter/VRPplotter.py