Invisible for both Camera and LiDAR: Security of Multi-Sensor Fusion based Perception in Autonomous Driving Under Physical-World Attacks
Open accessThe DOI matches its Crossref record; 29 of the 29 references it cites were found in a registry or an index.
Abstract
In Autonomous Driving (AD) systems, perception is both security and safety critical. Despite various prior studies on its security issues, all of them only consider attacks on camera-or LiDAR-based AD perception alone. However, production AD systems today predominantly adopt a Multi-Sensor Fusion (MSF) based design, which in principle can be more robust against these attacks under the assumption that not all fusion sources are (or can be) attacked at the same time. In this paper, we present the first study of security issues of MSF-based perception in AD systems. We directly challenge the basic MSF design assumption above by exploring the possibility of attacking all fusion sources simultaneously. This allows us for the first time to understand how much security guarantee MSF can fundamentally provide as a general defense strategy for AD perception.We formulate the attack as an optimization problem to generate a physically-realizable, adversarial 3D-printed object that misleads an AD system to fail in detecting it and thus crash into it. To systematically generate such a physical-world attack, we propose a novel attack pipeline that addresses two main design challenges: (1) non-differentiable target camera and LiDAR sensing systems, and (2) non-differentiable cell-level aggregated features popularly used in LiDAR-based AD perception. We evaluate our attack on MSF algorithms included in representative open-source industry-grade AD systems in real-world driving scenarios. Our results show that the attack achieves over 90% success rate across different object types and MSF algorithms. Our attack is also found stealthy, robust to victim positions, transferable across MSF algorithms, and physical-world realizable after being 3D-printed and captured by LiDAR and camera devices. To concretely assess the end-to-end safety impact, we further perform simulation evaluation and show that it can cause a 100% vehicle collision rate for an industry-grade AD system. We also evaluate and discuss defense strategies.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
The unreasonable effectiveness of deep features as a perceptual metric
Zhang, R., Isola, P., Efros, A. A., Shechtman, E., & Wang, O. (2018) · arXiv · doi:10.48550/arxiv.1801.03924 · Cited by 24,812 · cited 2×
“For LPIPS, we use the official implementation from [82] to measure the LPIPS value between the driving image with benign object rendered and the same image with the adversarial one rendered at the same location.”
MethodThe Unreasonable Effectiveness of Deep Features as a Perceptual Metric
Zhang, R., Isola, P., Efros, A. A., Shechtman, E., & Wang, O. (2018) · 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2018.00068 · Cited by 14,623
An Introduction to Genetic Algorithms
Mitchell, M. (1996) · doi:10.7551/mitpress/3927.001.0001 · Cited by 11,340
Explaining and Harnessing Adversarial Examples
Goodfellow, I. J., Shlens, J., & Szegedy, C. (2014) · arXiv · doi:10.48550/arxiv.1412.6572 · Cited by 10,381
Vision meets robotics: The KITTI dataset
Geiger, A., Lenz, P., Stiller, C., & Urtasun, R. (2013) · The International Journal of Robotics Research · doi:10.1177/0278364913491297 · Cited by 9,913 · cited 3×
“For each object type, we select 100 real-world driving scenarios from the KITTI dataset [36] in which such object in benign case can be 100% detected by the MSF combinations.”
MethodInfluentialTowards Evaluating the Robustness of Neural Networks
Carlini, N., & Wagner, D. (2017) · 2017 IEEE Symposium on Security and Privacy (SP) · doi:10.1109/sp.2017.49 · Cited by 8,859
Intriguing properties of neural networks
Szegedy, C., Zaremba, W., Sutskever, I., Bruna, J., Erhan, D., Goodfellow, I., & Fergus, R. (2013) · arXiv · doi:10.48550/arxiv.1312.6199 · Cited by 8,143
Towards Deep Learning Models Resistant to Adversarial Attacks
Madry, A., Makelov, A., Schmidt, L., Tsipras, D., & Vladu, A. (2017) · arXiv · doi:10.48550/arxiv.1706.06083 · Cited by 6,602
Object Detection With Deep Learning: A Review
Zhao, Z.-Q., Zheng, P., Xu, S.-T., & Wu, X. (2019) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2018.2876865 · Cited by 5,545 · cited 1×
“In the MSF algorithm, state-of-the-art designs predominantly adopt DNN networks to process the LiDAR-side and camera-side MSF inputs [15], [16], [18]–[24], [26], due to the recent superior performance of deep learning in object detection [44].”
MethodDeepFool: A Simple and Accurate Method to Fool Deep Neural Networks
Moosavi-Dezfooli, S.-M., Fawzi, A., & Frossard, P. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.282 · Cited by 5,317 · cited 1×
“Various adversarial attacks have been proposed to generate adversarial attacks in the digital space [12], [50]–[56], [66], [85], [105], [124]–[127].”
BackgroundVoxelNet: End-to-End Learning for Point Cloud Based 3D Object Detection
Zhou, Y., & Tuzel, O. (2018) · 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2018.00472 · Cited by 4,813
PointPillars: Fast Encoders for Object Detection From Point Clouds
Lang, A. H., Vora, S., Caesar, H., Zhou, L., Yang, J., & Beijbom, O. (2019) · 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2019.01298 · Cited by 4,673
The Limitations of Deep Learning in Adversarial Settings
Papernot, N., McDaniel, P., Jha, S., Fredrikson, M., Celik, Z. B., & Swami, A. (2016) · 2016 IEEE European Symposium on Security and Privacy (EuroS&P) · doi:10.1109/eurosp.2016.36 · Cited by 3,993
Multi-view 3D Object Detection Network for Autonomous Driving
Chen, X., Ma, H., Wan, J., Li, B., & Xia, T. (2017) · 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2017.691 · Cited by 3,450
Self-citationRobust physical-world attacks on deep learning visual classification
Eykholt, K., Evtimov, I., Fernandes, E., Li, B., Rahmati, A., Xiao, C., Prakash, A., Kohno, T., & Song, D. (2018) · 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2018.00175 · Cited by 2,173
Self-citationA Survey of Autonomous Driving: Common Practices and Emerging Technologies
Yurtsever, E., Lambert, J., Carballo, A., & Takeda, K. (2020) · IEEE Access · doi:10.1109/access.2020.2983149 · Cited by 1,889
Joint 3D Proposal Generation and Object Detection from View Aggregation
Ku, J., Mozifian, M., Lee, J., Harakeh, A., & Waslander, S. L. (2018) · 2018 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2018.8594049 · Cited by 1,697
PIXOR: Real-time 3D Object Detection from Point Clouds
Yang, B., Luo, W., & Urtasun, R. (2018) · 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2018.00798 · Cited by 1,249 · cited 4×
“…To improve run-time performance, the state-of-the-art LiDAR-based AD perception uses aggregated features of the 3D points grouped at the level of 2D or 3D cells [15], [20], [30]– [34]; however, their calculation is by nature non-differentiable (§III-B), which makes the attack difficult to optimize.”
BackgroundInfluentialKato, H., Ushiku, Y., & Harada, T. (2018) · 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2018.00411 · Cited by 954
PointFusion: Deep Sensor Fusion for 3D Bounding Box Estimation
Xu, D., Anguelov, D., & Jain, A. (2018) · 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2018.00033 · Cited by 856
Potential cyberattacks on automated vehicles
Petit, J., & Shladover, S. E. (2014) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2014.2342271 · Cited by 779
Multi-Task Multi-Sensor Fusion for 3D Object Detection
Liang, M., Yang, B., Chen, Y., Hu, R., & Urtasun, R. (2019) · 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2019.00752 · Cited by 723
Shafahi, A., Najibi, M., Ghiasi, A., Xu, Z., Dickerson, J., Studer, C., Davis, L. S., Taylor, G., & Goldstein, T. (2019) · ETH Zurich · doi:10.3929/ethz-b-000460345 · Cited by 580
Brown, T. B., Mané, D., Roy, A., Abadi, M., & Gilmer, J. (2017) · arXiv · doi:10.48550/arxiv.1712.09665 · Cited by 509
Voting for Voting in Online Point Cloud Object Detection
Zeng Wang, D., & Posner, I. (2015) · Robotics: Science and Systems XI · doi:10.15607/rss.2015.xi.035 · Cited by 337 · cited 3×
“Specifically, we compute Cell-level Aggregated Features Used in Occupancy [15], [16], [30], [34], [49], [73] Count [15], [16] Height (min/max/mean) [15], [16], [20], [24], [33] Intensity (min/max/mean) [15], [16], [20], [30], [33], [34], [49] Density [20], [24], [33] Table I. Summary of…”
BackgroundAdversarial Robustness: From Self-Supervised Pre-Training to Fine-Tuning
Chen, T., Liu, S., Chang, S., Cheng, Y., Amini, L., & Wang, Z. (2020) · 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr42600.2020.00078 · Cited by 205
Deep Steering: Learning End-to-End Driving Model from Spatial and Temporal Visual Cues
Chi, L., & Mu, Y. (2017) · arXiv · doi:10.48550/arxiv.1708.03798 · Cited by 87
BirdNet: A 3D Object Detection Framework from LiDAR Information
Beltran, J., Guindel, C., Moreno, F. M., Cruzado, D., Garcia, F., & De La Escalera, A. (2018) · 2018 21st International Conference on Intelligent Transportation Systems (ITSC) · doi:10.1109/itsc.2018.8569311 · Cited by 37
Accurate Monocular Object Detection via Color-Embedded 3D Reconstruction for Autonomous Driving
Ma, X., Wang, Z., Li, H., Zhang, P., Fan, X., & Ouyang, W. (2019) · arXiv · doi:10.48550/arxiv.1903.11444 · Cited by 12
Cited by
3D Object Detection for Autonomous Driving: A Comprehensive Survey
Mao, J., Shi, S., Wang, X., & Li, H. (2023) · International Journal of Computer Vision · doi:10.1007/s11263-023-01790-1 · Cited by 450
Visual Adversarial Examples Jailbreak Aligned Large Language Models
Qi, X., Huang, K., Panda, A., Henderson, P., Wang, M., & Mittal, P. (2024) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v38i19.30150 · Cited by 150
Metadata
- Authors
- Cao, Y., Wang, N., Xiao, C., Yang, D., Fang, J., Yang, R., Chen, Q. A., Liu, M., & Li, B.
- Venue
- 2021 IEEE Symposium on Security and Privacy (SP)
- Publisher
- IEEE
- Published
- 2021-05
- Pages
- 176-194
- ISSN
- 1081-6011 (ISSN-L)
- OpenAlex
- W3156877806
- Cited by
- 230
- References
- 29
- FWCI
- 18.14
- Type
- conference paper
- Language
- en
- Open access
- green
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.