Autonomous Vehicle Collision Avoidance Simulation

This project implements a collision avoidance simulation framework for autonomous vehicles using Model Predictive Control (MPC) and vehicle dynamics. The simulation framework is based on the methodology described in the reference paper:

Reference Paper:
Title: A MPC COMBINED DECISION MAKING AND TRAJECTORY PLANNING FOR AUTONOMOUS VEHICHLE COLLISION AVOIDANCE
Authors: Manel Ammour, Rodolfo Orjuela, and Michel Basset
Published in: IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS, VOL. 23, NO. 12 DECEMBER 2022

The project integrates several modules:

• Vehicle Dynamics: Implements the point-mass model for vehicles.
• MPC Controller: Formulates the MPC optimization problem using CasADi to compute control actions (accelerations & steering) to avoid collisions.
• Simulation: Manages the simulation loop, state updates, and logging of simulation data.
• Environment & Visualization: Contains the road model, lane definitions, and visualization utilities to plot vehicle trajectories and safety barriers.
• Scenarios: Sets up different driving scenarios (e.g., overtaking, lane changes) as described in Table II of the reference paper.
• Plotting Results: Utilities to visualize simulation data (e.g., trajectories, velocities, accelerations).

Directory Structure

Project_Root/
├── src/
│   ├── Env.py
│   ├── Fsm.py
│   ├── main.py
│   ├── Mpc_Controller.py
│   ├── Plot_Results.py
│   ├── Scenarios.py
│   ├── Simulation.py
│   ├── Utils.py
│   ├── Vehicle_Dynamics.py
│   ├── simulation_data.npz  # generated after simulation run
├── README.md
├── .gitignore
└── requirements.txt

Prerequisites

• Python 3.8+
• Recommended package dependencies are listed in the corresponding requirements file.

Key Dependencies

• NumPy
• Matplotlib
• CasADi

Setup and Installation

It is recommended to run the project in a Python virtual environment. Follow the steps below:

1. Clone the repository:

   git clone <repository_url>
   cd <repository_directory>

2. Create a virtual environment:

   On Windows:

   python -m venv venv
   venv\Scripts\activate

   On macOS/Linux:

   python3 -m venv venv
   source venv/bin/activate

3. Install dependencies:

   pip install -r requirements.txt

Compile and Run Instructions

Running the Simulation

After setting up the virtual environment and installing dependencies, run the simulation by executing:

python src/main.py

The simulation will:

• Initialize the environment and scenarios using parameters from Scenarios.py.
• Compute vehicle dynamics and control actions using the MPC in Mpc_Controller.py.
• Log simulation data and update vehicle states in Simulation.py.
• Visualize trajectories and safety barriers using the functions in Env.py.

Plotting Results

To generate graphical plots of the simulation logs, run:

python src/Plot_Results.py

This will load the simulation data from simulation_data.npz and produce plots such as trajectory charts, velocity profiles, and acceleration profiles.

Project Overview

This simulation framework aims to demonstrate collision avoidance capabilities through:

• MPC-based control: Using CasADi, the optimization problem includes acceleration and jerk constraints, ensuring smooth vehicle maneuvers.
• Dynamic Environment: Realistic vehicle interactions are modeled using a point-mass model, with lane information and road boundaries visualized.
• Scenario-based Testing: The system can switch between multiple driving scenarios (as defined in Scenarios.py) to evaluate performance under diverse conditions.

Further details, including mathematical formulations and experimental results, can be found in the reference paper provided.

Future Work

• Improve model fidelity by integrating more complex vehicle dynamics.
• Extend scenario diversity and include multi-agent interactions.
• Experiment with risk-averse formulations and further tuning of control parameters.

Acknowledgments

This project was developed as part of research in autonomous vehicle collision avoidance. Special thanks to Prof. Pradumn Kumar Pandey.

---

For any questions or issues, please create an issue in this repository or contact the project maintainer here.