SoK: The Next Frontier in AV Security: Systematizing Perception Attacks and the Emerging Threat of Multi-Sensor Fusion

2026-04-22

AI Summary

This work addresses the growing threat of cross-modal attacks against multi-sensor fusion perception systems in autonomous driving, a domain where existing research lacks systematic analysis of fusion logic vulnerabilities and effective defenses. The study presents the first comprehensive attack classification framework tailored to multi-sensor fusion, grounded in a Systematization of Knowledge (SoK) review of 48 perception-layer attack studies. It establishes a unified taxonomy encompassing sensor modality, attack stage, medium, and perception module. Through simulated joint spoofing experiments involving infrared and LiDAR sensors, the research uncovers critical security blind spots within fusion hierarchies. Notably, it reveals that sensor redundancy—a cornerstone of robust perception—can itself be exploited as an attack vector, highlighting deficiencies in real-world validation and long-term evaluation in current approaches, and thereby charting a new direction toward trustworthy fusion-based defense mechanisms.

Technology Category

• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Computer Vision: Multi-modal Vision
• Machine Learning: Multimodal Learning

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Autonomous vehicles (AVs) increasingly rely on multi-sensor perception pipelines that combine data from cameras, lidar, radar, and other modalities to interpret the environment. This SoK systematizes 48 peer-reviewed studies on perception-layer attacks against AVs, tracking the field's evolution from single-sensor exploits to complex cross-modal threats that compromise multi-sensor fusion (MSF). We develop a unified taxonomy of 20 attack vectors organized by sensor type, attack stage, medium, and perception module, revealing patterns that expose underexplored vulnerabilities in fusion logic and cross-sensor dependencies. Our analysis identifies key research gaps, including limited real-world testing, short-term evaluation bias, and the absence of defenses that account for inter-sensor consistency. To illustrate one such gap, we validate a fusion-level vulnerability through a proof-of-concept simulation combining infrared and lidar spoofing. The findings highlight a fundamental shift in AV security: as systems fuse more sensors for robustness, attackers exploit the very redundancy meant to ensure safety. We conclude with directions for fusion-aware defense design and a research agenda for trustworthy perception in autonomous systems.

Problem

• perception attacks
• multi-sensor fusion
• autonomous vehicles
• AV security
• sensor spoofing

Innovation

• multi-sensor fusion
• perception attacks
• autonomous vehicle security
• attack taxonomy
• cross-modal threats