Back to projects
Energy Efficient Vehicle Routing
STAT 4830 Project, Mathematical Optimizations for Machine Learning and Data Science
2025Graph Neural NetworksReinforcement LearningTransformer Models
Three vehicle routing algorithms to minimize energy consumption for vehicles
Tech Stack
PythonPandasNumPyPyTorchJupyter
Project Description
- In a team of three, I worked to develop three vehicle routing algorithms that optimize on travel time and energy consumption.
- Used a dataset of 300+ vehicles and their respective routes marked by GPS coordinates.
- Trained all of our models in Jupyter notebooks using PyTorch library on Pandas dataframes
Key Features
- Designed and implemented Graph Neural Network, BERT-based transformer, and Deep Q-network models to predict optimal vehicle routing
- Implemented a custom loss function that penalizes for longer routes and higher energy consumption.
- Considered geographical features such as elevation and road features such as speed limits and congestion
- Used AdamW optimizer for training.
Preview
