🤖 AI Summary
This work addresses physical adversarial examples induced by vehicular camera hardware failures—specifically, optical component damage such as lens glass fracture. We propose the first modeling framework formalizing optical hardware failure as a novel class of physical adversarial attacks. Methodologically, we develop a physics-based simulation model of damaged lenses grounded in stress propagation mechanics, integrating constrained particle grids with stochastic stress injection to synthesize realistic image distortions. We further introduce a graph-structured constrained propagation neural simulator that efficiently learns the nonlinear stress-to-image mapping. Experiments demonstrate that generated samples significantly increase target detection failure rates; moreover, their Fréchet Inception Distance closely matches that of real-world damaged images, validating both simulation fidelity and adversarial effectiveness. This work establishes the first scalable, physics-informed simulation paradigm for evaluating autonomous driving robustness against optical hardware failures.
📝 Abstract
While there has been extensive work on generating physics-based adversarial samples recently, an overlooked class of such samples come from physical failures in the camera. Camera failures can occur as a result of an external physical process, i.e. breakdown of a component due to stress, or an internal component failure. In this work, we develop a simulated physical process for generating broken lens as a class of physics-based adversarial samples. We create a stress-based physical simulation by generating particles constrained in a mesh and apply stress at a random point and at a random angle. We perform stress propagation through the mesh and the end result of the mesh is a corresponding image which simulates the broken lens pattern. We also develop a neural emulator which learns the non-linear mapping between the mesh as a graph and the stress propagation using constrained propagation setup. We can then statistically compare the difference between the generated adversarial samples with real, simulated and emulated adversarial examples using the detection failure rate of the different classes and in between the samples using the Frechet Inception distance. Our goal through this work is to provide a robust physics based process for generating adversarial samples.