Home
AI50 project to modelize Capacited Vehicle Routing Problem (CVRP) and find solutions using Multi-Agents Systems, metaheuristics and Machine Learning.
This project requires a 3.8+ python version and multiple library to be runned. Before running it for the first time please ensure that you have every python library required by running the following command:
pip install requirements.txtTo launch the web application, run the main.py file. You can add one or more of the followings arguments:
| Argument | Explanation |
|---|---|
| -h or --help | Display the help message |
| -t or --unittest | Run unit test before running the application |
| -s or --show_evolution | Display the current solution on the load page |
The web application runned by default on https://localhost:8080/. To be runned the application needs Redis server available (version for windows and linux are present in the redis folder).
The algorithm is based on the scientific article Agents toward Vehicle Routing Problem. It is written in sarl and bridged to python using py4J library.
Two metaheuristics have been implemented:
- Tabu search
- Genetic Algorithm
Both metaheuristics start from a solution provided by the Clark & Wright saving algorithm and try to improve it. Here's a small example of the result provided by the tabu search algorithm.
