Optimizing Vehicle Routing with Graph-Based and Probabilistic Models

Team Members

Vishva Gajaraj, Darren Mo, Aman Sharma

High-Level Summary

This project tackles the challenge of optimizing vehicle routing by developing algorithms that minimize both travel time and energy consumption. We've created a comprehensive approach that provides more detailed insights than current routing applications by considering trip-specific features, geographic data, infrastructure elements, and traffic patterns.

Our key contributions include:

• A systematic pipeline for modeling travel behavior and energy usage
• Benchmark statistics comparing different routing models

Repository Structure

├── src/                  # Source code for final models and algorithms
├── notebooks/            # Jupyter notebooks for analysis and demonstrations
├── docs/                 # Documentation files
├── report.md             # Comprehensive project report with methodology and findings
├── requirements.txt      # Python dependencies
└── named_development_history/ # Archive of development process and iterations

The named_development_history/ folder contains our journey through model exploration and development, preserved for reference purposes. However, the final, optimized code exists in the src directory.

Setup Instructions

Prerequisites

• Python 3.8 or higher
• Git

Environment Setup

1. Clone the repository:

   git clone [repository-url]
   cd [repository-name]
   

2. Create and activate a virtual environment (recommended):

   python -m venv venv
   source venv/bin/activate  # On Windows: venv\Scripts\activate
   

3. Install dependencies:

   pip install -r requirements.txt
   

4. API Keys (if needed):

   • OpenStreetMap API is used but doesn't require authentication

5. Running Code

   • The Code works in isolation as long as the data is stored locally. The link to the dataset is attached below.
   • https://bitbucket.org/datarepo/eved-dataset/src/main/

Demo Notebook

We left out the data processing and model training and simply inputted a saved model.pth file that can be downloaded from within our repo (for the sake of computational resources for the reproducer).

This notebook provides a step-by-step walkthrough of our approach of benchmarking our key successful models against the deterministic approaches of models like A* and Djikstra.

[Interactive Demo] https://drive.google.com/file/d/1TLNbLeSUoYlhsLNXXKqOkOlkR76TOOoM/view?usp=sharing