Runtime Stealthy Perception Attacks against DNN-based Adaptive Cruise Control Systems

2023-07-18

AI Summary

This study exposes a stealthy camera-based perception attack against deep neural network–driven Adaptive Cruise Control (ACC) systems: strategically crafted image perturbations induce erroneous leading-vehicle distance estimation, escalating rear-end collision risk. To address the lack of context awareness and physical realizability in existing attacks, we propose a context-aware attack timing selection strategy and a real-time adaptive perturbation optimization method—ensuring stealthiness, robustness, and high physical-world success. Our evaluation integrates real-vehicle testing, high-fidelity simulation (incorporating production-grade ACC and Automatic Emergency Braking Systems), public benchmarks, and human-in-the-loop intervention assessment. Results demonstrate a 142.9× improvement in attack success rate over baseline methods, an 82.6% increase in evasion rate against detection, and—crucially—the first quantitative validation of human drivers’ and baseline safety mechanisms’ critical defensive roles.

Abstract

Adaptive Cruise Control (ACC) is a widely used driver assistance technology for maintaining the desired speed and safe distance to the leading vehicle. This paper evaluates the security of the deep neural network (DNN) based ACC systems under runtime stealthy perception attacks that strategically inject perturbations into camera data to cause forward collisions. We present a context-aware strategy for the selection of the most critical times for triggering the attacks and a novel optimization-based method for the adaptive generation of image perturbations at runtime. We evaluate the effectiveness of the proposed attack using an actual vehicle, a publicly available driving dataset, and a realistic simulation platform with the control software from a production ACC system, a physical-world driving simulator, and interventions by the human driver and safety features such as Advanced Emergency Braking System (AEBS). Experimental results show that the proposed attack achieves 142.9 times higher success rate in causing hazards and 82.6% higher evasion rate than baselines, while being stealthy and robust to real-world factors and dynamic changes in the environment. This study highlights the role of human drivers and basic safety mechanisms in preventing attacks.

Problem

• Adaptive Cruise Control
• Covert Attack
• Camera Interference

Innovation

• Adaptive Cruise Control Attack
• Camera Image Manipulation
• Stealthy Attack Methodology