Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
97 tokens/sec
GPT-4o
53 tokens/sec
Gemini 2.5 Pro Pro
44 tokens/sec
o3 Pro
5 tokens/sec
GPT-4.1 Pro
47 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

Generating Transferable Adversarial Simulation Scenarios for Self-Driving via Neural Rendering (2309.15770v3)

Published 27 Sep 2023 in cs.RO

Abstract: Self-driving software pipelines include components that are learned from a significant number of training examples, yet it remains challenging to evaluate the overall system's safety and generalization performance. Together with scaling up the real-world deployment of autonomous vehicles, it is of critical importance to automatically find simulation scenarios where the driving policies will fail. We propose a method that efficiently generates adversarial simulation scenarios for autonomous driving by solving an optimal control problem that aims to maximally perturb the policy from its nominal trajectory. Given an image-based driving policy, we show that we can inject new objects in a neural rendering representation of the deployment scene, and optimize their texture in order to generate adversarial sensor inputs to the policy. We demonstrate that adversarial scenarios discovered purely in the neural renderer (surrogate scene) can often be successfully transferred to the deployment scene, without further optimization. We demonstrate this transfer occurs both in simulated and real environments, provided the learned surrogate scene is sufficiently close to the deployment scene.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (67)
  1. K. Nidhi and S. M. Paddock. Driving to Safety: How Many Miles of Driving Would It Take to Demonstrate Autonomous Vehicle Reliability? https://www.rand.org/pubs/research_reports/RR1478.html, 2016. [Online; accessed 19-July-2018].
  2. A. C. Madrigal. Inside waymo’s secret world for training self-driving cars. The Atlantic, Aug 2017. URL https://www.theatlantic.com/technology/archive/2017/08/inside-waymos-secret-testing-and-simulation-facilities/537648/.
  3. Exploring adversarial robustness of multi-sensor perception systems in self driving, Jan. 2022. URL https://arxiv.org/abs/2101.06784.
  4. Physically Realizable Adversarial Examples for LiDAR Object Detection, Apr. 2020. URL https://arxiv.org/abs/2004.00543.
  5. Finding Physical Adversarial Examples for Autonomous Driving with Fast and Differentiable Image Compositing, June 2021. URL https://arxiv.org/abs/2010.08844.
  6. Adversarial objects against lidar-based autonomous driving systems. CoRR, abs/1907.05418, 2019. URL http://arxiv.org/abs/1907.05418.
  7. Drivegan: Towards a controllable high-quality neural simulation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 5820–5829, June 2021.
  8. Geosim: Realistic video simulation via geometry-aware composition for self-driving. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 7230–7240, June 2021.
  9. Unisim: A neural closed-loop sensor simulator. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 1389–1399, June 2023.
  10. AdvSim: Generating Safety-Critical Scenarios for Self-Driving Vehicles, Jan. 2022. URL https://arxiv.org/abs/2101.06549.
  11. Generating useful accident-prone driving scenarios via a learned traffic prior. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 17305–17315, 2022.
  12. Symphony: Learning realistic and diverse agents for autonomous driving simulation, 2022. URL https://arxiv.org/abs/2205.03195.
  13. Generating adversarial driving scenarios in high-fidelity simulators. In 2019 International Conference on Robotics and Automation (ICRA), pages 8271–8277. IEEE, 2019.
  14. Overriding Autonomous Driving Systems Using Adaptive Adversarial Billboards. IEEE Transactions on Intelligent Transportation Systems, 23(8):11386–11396, Aug. 2022.
  15. Vista 2.0: An open, data-driven simulator for multimodal sensing and policy learning for autonomous vehicles. In 2022 International Conference on Robotics and Automation (ICRA), pages 2419–2426, 2022.
  16. Adaptive stress testing for autonomous vehicles. In 2018 IEEE Intelligent Vehicles Symposium (IV), pages 1–7. IEEE, 2018.
  17. C. Innes and S. Ramamoorthy. Testing rare downstream safety violations via upstream adaptive sampling of perception error models. In 2023 IEEE International Conference on Robotics and Automation (ICRA), pages 12744–12750. IEEE, 2023.
  18. Advances in adversarial attacks and defenses in computer vision: A survey, Sept. 2021. URL https://arxiv.org/abs/2108.00401.
  19. Synthesizing robust adversarial examples. In Proceedings of the 35th International Conference on Machine Learning, volume 80, pages 284–293, 2018.
  20. Targeted attack on deep rl-based autonomous driving with learned visual patterns. In 2022 International Conference on Robotics and Automation (ICRA), pages 10571–10577, 2022.
  21. Advances in neural rendering. In ACM SIGGRAPH 2021 Courses, SIGGRAPH ’21, 2021.
  22. Nerf2real: Sim2real transfer of vision-guided bipedal motion skills using neural radiance fields, 2022. URL https://arxiv.org/abs/2210.04932.
  23. Plenoxels: Radiance fields without neural networks. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 5501–5510, 2022.
  24. Neural sparse voxel fields. Advances in Neural Information Processing Systems, 33:15651–15663, 2020.
  25. Tensorf: Tensorial radiance fields. European Conference on Computer Vision (ECCV), 2022.
  26. Instant neural graphics primitives with a multiresolution hash encoding. ACM Transactions on Graphics, 41(4):1–15, July 2022.
  27. King: Generating safety-critical driving scenarios for robust imitation via kinematics gradients, 2022. URL https://arxiv.org/abs/2204.13683.
  28. L. S. Pontryagin. Mathematical theory of optimal processes. CRC press, 1987.
  29. PyTorch: An Imperative Style, High-Performance Deep Learning Library. In H. Wallach, H. Larochelle, A. Beygelzimer, F. d’Alché Buc, E. Fox, and R. Garnett, editors, Advances in Neural Information Processing Systems 32, pages 8024–8035. Curran Associates, Inc., 2019.
  30. gradsim: Differentiable simulation for system identification and visuomotor control. International Conference on Learning Representations (ICLR), 2021. URL https://openreview.net/forum?id=c_E8kFWfhp0.
  31. CARLA: An open urban driving simulator. In Proceedings of the 1st Annual Conference on Robot Learning, pages 1–16, 2017.
  32. End-to-end interpretable neural motion planner. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 8660–8669, 2019.
  33. Transfuser: Imitation with transformer-based sensor fusion for autonomous driving. Pattern Analysis and Machine Intelligence (PAMI), 2022.
  34. A survey of end-to-end driving: Architectures and training methods. IEEE Transactions on Neural Networks and Learning Systems, 33(4):1364–1384, 2020.
  35. Block-nerf: Scalable large scene neural view synthesis. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 8248–8258, 2022.
  36. S-neRF: Neural radiance fields for street views. In International Conference on Learning Representations, 2023.
  37. Nerf-vae: A geometry aware 3d scene generative model, 2021. URL https://arxiv.org/abs/2104.00587.
  38. M. Niemeyer and A. Geiger. Giraffe: Representing scenes as compositional generative neural feature fields. In Proc. IEEE Conf. on Computer Vision and Pattern Recognition (CVPR), 2021.
  39. Learning object-compositional neural radiance field for editable scene rendering. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 13779–13788, 2021.
  40. Volumetric disentanglement for 3d scene manipulation, 2022. URL https://arxiv.org/abs/2206.02776.
  41. Spin-nerf: Multiview segmentation and perceptual inpainting with neural radiance fields, 2022. URL https://arxiv.org/abs/2211.12254.
  42. Nerv: Neural reflectance and visibility fields for relighting and view synthesis. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 7495–7504, 2021.
  43. Intrinsicnerf: Learning intrinsic neural radiance fields for editable novel view synthesis, 2022. URL https://arxiv.org/abs/2210.00647.
  44. Discoscene: Spatially disentangled generative radiance field for controllable 3d-aware scene synthesis, 2022. URL https://arxiv.org/abs/2212.11984.
  45. Panoptic neural fields: A semantic object-aware neural scene representation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 12871–12881, 2022.
  46. Inerf: Inverting neural radiance fields for pose estimation. In 2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), page 1323–1330, 2021.
  47. Vision-only robot navigation in a neural radiance world. IEEE Robotics and Automation Letters, 7(2):4606–4613, 2022.
  48. Differentiable physics simulation of dynamics-augmented neural objects, 2022. URL https://arxiv.org/abs/2210.09420.
  49. Learning multi-object dynamics with compositional neural radiance fields, 2022. URL https://arxiv.org/abs/2202.11855.
  50. Objaverse: A universe of annotated 3d objects, 2022. URL https://arxiv.org/abs/2212.08051.
  51. B. O. Community. Blender - a 3D modelling and rendering package. Blender Foundation, Stichting Blender Foundation, Amsterdam, 2018. URL http://www.blender.org.
  52. End to end learning for self-driving cars, 2016. URL https://arxiv.org/abs/1604.07316.
  53. A reduction of imitation learning and structured prediction to no-regret online learning. In Proceedings of the fourteenth international conference on artificial intelligence and statistics, pages 627–635. JMLR Workshop and Conference Proceedings, 2011.
  54. End-to-end driving via conditional imitation learning. CoRR, abs/1710.02410, 2017. URL http://arxiv.org/abs/1710.02410.
  55. Structure-from-motion revisited. In Conference on Computer Vision and Pattern Recognition (CVPR), 2016.
  56. Pixelwise view selection for unstructured multi-view stereo. In European Conference on Computer Vision (ECCV), 2016.
  57. Accelerating 3d deep learning with pytorch3d. arXiv:2007.08501, 2020.
  58. Large steps in inverse rendering of geometry. ACM Transactions on Graphics (Proceedings of SIGGRAPH Asia), 40(6), Dec. 2021. doi:10.1145/3478513.3480501.
  59. Black-box optimization with local generative surrogates. Advances in Neural Information Processing Systems, 33:14650–14662, 2020.
  60. Bayesian optimization in high dimensions via random embeddings. In IJCAI, pages 1778–1784, 2013.
  61. Bayesian optimization with gradients, 2018.
  62. Do differentiable simulators give better policy gradients?, 2022.
  63. Unbiased gradient estimation in unrolled computation graphs with persistent evolution strategies, 2021.
  64. Gradient estimation using stochastic computation graphs. Advances in neural information processing systems, 28, 2015.
  65. R. Astudillo and P. I. Frazier. Thinking inside the box: A tutorial on grey-box bayesian optimization. In 2021 Winter Simulation Conference (WSC), pages 1–15. IEEE, 2021.
  66. Differentiation of blackbox combinatorial solvers. arXiv preprint arXiv:1912.02175, 2019.
  67. Differentiable convex optimization layers. Advances in neural information processing systems, 32, 2019.
User Edit Pencil Streamline Icon: https://streamlinehq.com
Authors (7)
  1. Yasasa Abeysirigoonawardena (1 paper)
  2. Kevin Xie (13 papers)
  3. Chuhan Chen (2 papers)
  4. Salar Hosseini (1 paper)
  5. Ruiting Chen (1 paper)
  6. Ruiqi Wang (62 papers)
  7. Florian Shkurti (52 papers)
Citations (1)
View on Semantic Scholar

Summary

We haven't generated a summary for this paper yet.

Ai Sparkles Streamline Icon: https://streamlinehq.com Summarize Now