Exploring Adversarial Robustness of LiDAR Semantic Segmentation in Autonomous Driving
Open Access
Abstract
Deep learning networks have demonstrated outstanding performance in 2D and 3D vision tasks. However, recent research demonstrated that these networks result in failures when imperceptible perturbations are added to the input known as adversarial attacks. This phenomenon has recently received increased interest in the field of autonomous vehicles and has been extensively researched on 2D image-based perception tasks and 3D object detection. However, the adversarial robustness of 3D LiDAR semantic segmentation in autonomous vehicles is a relatively unexplored topic. This study expands the adversarial examples to LiDAR-based 3D semantic segmentation. We developed and analyzed three LiDAR point-based adversarial attack methods on different networks developed on the SemanticKITTI dataset. The findings illustrate that the Cylinder3D network has the highest adversarial susceptibility to the analyzed attacks. We investigated how the class-wise point distribution influences the adversarial robustness of each class in the SemanticKITTI dataset and discovered that ground-level points are extremely vulnerable to point perturbation attacks. Further, the transferability of each attack strategy was assessed, and we found that networks relying on point data representation demonstrate a notable level of resistance. Our findings will enable future research in developing more complex and specific adversarial attacks against LiDAR segmentation and countermeasures against adversarial attacks.Keywords
This publication has 31 references indexed in Scilit:
- Physically Realizable Adversarial Examples for LiDAR Object Detection2020
- PolarNet: An Improved Grid Representation for Online LiDAR Point Clouds Semantic Segmentation2020
- Adversarial Attacks and Defenses in Images, Graphs and Text: A ReviewInternational Journal of Automation and Computing, 2020
- Adversarial Sensor Attack on LiDAR-based Perception in Autonomous Driving2019
- Extending Adversarial Attacks and Defenses to Deep 3D Point Cloud Classifiers2019
- 4D Spatio-Temporal ConvNets: Minkowski Convolutional Neural Networks2019
- Generating 3D Adversarial Point Clouds2019
- On the Robustness of Semantic Segmentation Models to Adversarial Attacks2018
- Boosting Adversarial Attacks with Momentum2018
- Robust Physical-World Attacks on Deep Learning Visual Classification2018