Risk-Aware Autonomous Driving with Linear Temporal Logic Specifications

Shuhao Qi¹, Zengjie Zhang ¹, Zhiyong Sun² and Sofie Haesaert^1
1 Eindhoven University of Technology ² Peking University

Overview

This repository provides a minimal example of the risk-aware planning framework proposed in the paper: ArXiv:2409.09769, without relying on Carla. This example is designed to help users understand the proposed framework in a simplified setting.

Repository Structure

• reactive_LTL_main.py - Main script handling reactive Linear Temporal Logic (LTL) computations.

Abstraction Module (abstraction/)

• abstraction.py - Implements abstraction mechanisms for system modeling.
• MDP.py - Defines a Markov Decision Process (MDP) abstraction.
• prod_MDP.py - Implements a product MDP for combining different abstractions.

Risk-aware Linear Programming Module (risk_LP/)

• risk_LP.py - Risk-aware linear programming (LP) problems.
• ltl_risk_LP.py - Implements risk-aware LP methods under LTL constraints.
• prod_auto.py - Construct product automaton.
• risk_field_plot.py - Provides visualization tools for risk field data.

Simulation Module (sim/)

• simulator.py - Simulates the environment and vehicle dynamic.
• perception.py - Analog perception module to provide sensor information.
• visualizer.py - PLot simulation results.
• controller.py - Low-level MPC controller for the bicycle model.

Specification Module (specification/)

• specification.py - Translate LTL specifications.
• DFA.py - Define deterministic finite automata (DFA).

Getting Started

Prerequisites

Ensure you have the following installed:

• Python 3.x
• Required dependencies: Gurobi, numpy, matplotlib, networkx, and scipy.

Quick Start

Run the main script:

python reactive_LTL_main.py

License

This project is licensed under the MIT License. Please consider citing our papers if the project helps your research with the following BibTex:

@INPROCEEDINGS{qi2025risk,
  author={Qi, Shuhao and Zhang, Zengjie and Sun, Zhiyong and Haesaert, Sofie},
  booktitle={2025 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)}, 
  title={Risk-Aware Autonomous Driving with Linear Temporal Logic Specifications}, 
  year={2025},
  pages={14877-14883}}