Adversarial Robustness of Deep Sensor Fusion Models

TL;DR

This paper investigates the robustness of deep sensor fusion models combining camera and LiDAR data for autonomous driving, revealing insights into their accuracy, vulnerability, and effects of adversarial training strategies.

Contribution

It provides a comprehensive experimental analysis of adversarial robustness in deep sensor fusion models, highlighting the impact of fusion strategies and adversarial training methods.

Findings

Fusion models are more accurate and robust than single-sensor models.

Early fusion is more robust than late fusion without adversarial training.

Joint-channel adversarial training offers limited robustness improvements.

Abstract

We experimentally study the robustness of deep camera-LiDAR fusion architectures for 2D object detection in autonomous driving. First, we find that the fusion model is usually both more accurate, and more robust against single-source attacks than single-sensor deep neural networks. Furthermore, we show that without adversarial training, early fusion is more robust than late fusion, whereas the two perform similarly after adversarial training. However, we note that single-channel adversarial training of deep fusion is often detrimental even to robustness. Moreover, we observe cross-channel externalities, where single-channel adversarial training reduces robustness to attacks on the other channel. Additionally, we observe that the choice of adversarial model in adversarial training is critical: using attacks restricted to cars' bounding boxes is more effective in adversarial training and exhibits less significant cross-channel externalities. Finally, we find that joint-channel adversarial training helps mitigate many of the issues above, but does not significantly boost adversarial robustness.