Online Adaptive Platoon Control for Connected and Automated Vehicles via Physical Enhanced Residual Learning
This repo provides the source code and data for the following paper:
P. Zhang, H. Zhou, H. Huang, H. Shi, K. Long and X. Li, "Online Adaptive Platoon Control for Connected and Automated Vehicles via Physical Enhanced Residual Learning".
This paper introduces a physically enhanced residual learning (PERL) framework for connected and automated vehicle (CAV) platoon control, addressing the dynamics and unpredictability inherent to platoon systems. The framework combines a physical model of vehicle platoons with data-driven online learning methods, enhancing centralized platoon control and emphasizing multi-objective collaborative optimization. The residual controller, based on neural network (NN) learning, enriches the prior knowledge of the physical model and corrects residuals caused by vehicle dynamics.
The project is developed by Python 3. Please ensure you have a Python 3 environment set up.
The code related to our data processing and algorithm are:
- simulation/trajectory_generation.py - Code used to generate reference trajectories through IDM model or the OpenACC dataset.
- simulation/PERL_simulation.py - The main function calls the PERL algorithm.
- physical experiment/ - Code related to the reduced-scale platform.
As you proceed through each Python script, always verify the paths for both the input and output files. This ensures that everything runs smoothly.
Folder 'data' contains some sample trajectories with a brief readme.
Developer - Peng Zhang, Hang Zhou (pzhang257@wisc.edu, hzhou364@wisc.edu).
If you have any questions, please feel free to contact CATS Lab in UW-Madison. We're here to help!