Integrated Simulation Framework for Adversarial Attacks on Autonomous Vehicles

2025-08-31

AI Summary

Existing simulation frameworks lack support for evaluating multi-domain adversarial attacks targeting both perception and communication layers of autonomous vehicles (AVs). This paper introduces the first open-source integrated simulation framework enabling end-to-end modeling of joint adversarial scenarios—spanning LiDAR-based 3D object detection and V2X communication (e.g., message tampering, GPS spoofing). The framework unifies high-fidelity environment, traffic flow, and V2X network simulation within a ROS 2–based orchestration layer, allowing cross-domain attack generation via a single configuration file and seamless integration with mainstream AV software stacks. Experimental results demonstrate that the generated adversarial samples significantly degrade the performance of state-of-the-art 3D detectors, effectively exposing critical safety vulnerabilities of AV systems under realistic adversarial conditions.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Multiagent Systems: Adversarial Agents
• Application Domains: Mobility, Driving & Flight

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems

Abstract

Autonomous vehicles (AVs) rely on complex perception and communication systems, making them vulnerable to adversarial attacks that can compromise safety. While simulation offers a scalable and safe environment for robustness testing, existing frameworks typically lack comprehensive supportfor modeling multi-domain adversarial scenarios. This paper introduces a novel, open-source integrated simulation framework designed to generate adversarial attacks targeting both perception and communication layers of AVs. The framework provides high-fidelity modeling of physical environments, traffic dynamics, and V2X networking, orchestrating these components through a unified core that synchronizes multiple simulators based on a single configuration file. Our implementation supports diverse perception-level attacks on LiDAR sensor data, along with communication-level threats such as V2X message manipulation and GPS spoofing. Furthermore, ROS 2 integration ensures seamless compatibility with third-party AV software stacks. We demonstrate the framework's effectiveness by evaluating the impact of generated adversarial scenarios on a state-of-the-art 3D object detector, revealing significant performance degradation under realistic conditions.

Problem

• Simulating multi-domain adversarial attacks on autonomous vehicles
• Evaluating robustness of perception and communication systems
• Assessing safety impacts under realistic attack scenarios

Innovation

• Integrated simulation framework for multi-domain adversarial attacks
• Unified core synchronizes multiple simulators via single configuration
• Supports LiDAR attacks and V2X message manipulation threats