Mixed-Vehicle Platooning Control with a Gaussian Process Learning-Based Model

This repository contains the source code for research on enhancing safety in mixed-traffic platooning through a hybrid model. This approach integrates Gaussian Process (GP) learning with traditional first-principles modeling to predict behaviors of human-driven vehicles in mixed traffic scenarios.

Features

• Hybrid Modeling: Combines GP learning with first-principles models for accurate predictions.
• Real-Time Performance: Optimized for real-time applications with minimal computational load.
• Safety Enhancements: Demonstrates improved vehicle distances and platoon speeds in simulations.

This work is detailed in the paper:

• Enhancing Safety in Mixed Traffic: Learning-Based Modeling and Efficient Control of Autonomous and Human-Driven Vehicles

ARX_simulation

This directory contains code for simulating a vehicle platoon using Model Predictive Control (MPC) with an AutoRegressive with eXogenous inputs (ARX) model combined with a Gaussian Process (GP). It includes:

• MPC_platoon_simulation.m: Main simulation script.
• MPC_platoon.m: Implementation of the MPC.
• GP_sysmodel.m: Vehicle dynamics model for simulations.
• gpr_medium.mat: Standard GP model trained on field data.
• gpCallback.m: Callback function for CasADi to integrate the GP model (ARX+GP)
• wltp_velocity_profile.m: Script for the Worldwide Harmonized Light Vehicles Test Procedure (WLTP) velocity profile.

GP_ARX_simulation

This folder contains the code for simulating a platoon of vehicles using MPC with a sparseGP+ARX model, while the simulated vehicle is a GP+ARX model. Specifically, the following files are included:

• GP_MPC_platoon_simulation.m: Main script for sparseGP-MPC-based simulations.
• GP_MPC_platoon.m: Gaussian Process Model Predictive Control (GP-MPC) implementation.
• GP_sysmodel.m: Vehicle dynamics model for simulations.
• gpr_medium.mat: Standard GP model trained on field data.
• gpr_sparse.mat: Sparse GP model, trained using GP_RE_trainForSpares.m in the GP_training folder.
• gpCallback.m: Callback for CasADi, loading the sparseGP+ARX model for control purposes.
• gpCallback_sys.m: Callback for CasADi, integrating the GP+ARX model for simulating autonomous vehicles.
• wltp_velocity_profile.m: WLTP velocity profile script.

GP_training

Contains scripts for training Gaussian Process (GP) models:

• GP_training: Script for training standard GP models.
• GP_RE_trainForSpares: Script for developing sparse GP models.
• plot_to_pdf: Utility to export plot visualizations as PDFs.

Citing Our Work

If our work aids your research or you use the GP-MPC framework, please consider citing:

@article{wang2024enhancing,
  title={Enhancing safety in mixed traffic: Learning-based modeling and efficient control of autonomous and human-driven vehicles},
  author={Wang, Jie and Pant, Yash Vardhan and Zhao, Lei and Antkiewicz, Micha{\l} and Czarnecki, Krzysztof},
  journal={IEEE Transactions on Intelligent Transportation Systems},
  year={2024},
  publisher={IEEE}
}

Please also check our preliminary work at GP-MPC-of-Platooning.

@article{wang2024improving,
  title={Improving safety in mixed traffic: A learning-based model predictive control for autonomous and human-driven vehicle platooning},
  author={Wang, Jie and Jiang, Zhihao and Pant, Yash Vardhan},
  journal={Knowledge-Based Systems},
  volume={293},
  pages={111673},
  year={2024},
  publisher={Elsevier}
}

@article{wang2024learning,
  title={Learning-based modeling of human-autonomous vehicle interaction for improved safety in mixed-vehicle platooning control},
  author={Wang, Jie and Pant, Yash Vardhan and Jiang, Zhihao},
  journal={Transportation Research Part C: Emerging Technologies},
  volume={162},
  pages={104600},
  year={2024},
  publisher={Elsevier}
}