A Geometry and Semantics-Based Progressive Generation Approach for Traffic Sign Adversarial Patches

A general approach on the quantification of effective quantities in a pressure-driven flow of inelastic non-Newtonian fluid in arbitrary geometries based on the energy balance

The multiple source location problem is solved for a general array geometry using the principal of least-squares modeling. The incident wavefields may be a mixture of coherent and noncoherent sources that are broadband and/or narrowband. Furthermore, the wavefields may be plane waves and/or spherical waves. An effective algorithmic approach is developed for solving the multiple source location problem in the signal and covariance domains. In each domain, it is necessary to represent a set of vectors as a linear combination of array steering vectors. Using this property, a generic descent algorithmic approach is developed for estimating the wavefield directions of arrival. The success of such iterative approaches is critically dependent on the generation of good estimates to initialize the algorithm. An effective initialization procedure is developed for this purpose. Direction-of-arrival estimates which arise from the proposed procedure are empirically found to be unbiased and low variance.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Machining process productivity is adversely impacted by the unstable chatter vibrations causing poor surface quality, excessive loads, and premature tool and machine failures. Depending on the position of the cutting edge element (CEE) along the vibration wave on the part surface, its flank face and hone radius dynamically indent into the material and cause process damping (PD), which improves the machining stability. This article proposes a generalized analytical PD model by lifting the straight flank face limitation of authors’ previously developed approach and extends it to any clearance face geometries used in machining operations. The proposed model employs the dimensionality reduction method by discretizing the two-dimensional (2D) contact between the CEE and the part surface with series of springs to simulate the contact mechanics between the two. Elasto-plastic material model of the workpiece is used to calculate the contact pressure considering the work material properties, cutting edge geometry, machining and vibration parameters as inputs. The PD force is evaluated by removing the effect of the static indentation from the overall contact force. The equivalent viscous damping coefficient is calculated to linearize the PD force and can be used for accurate machining stability prediction of difficult-to-cut materials. The model has been validated with the experimentally identified and finite elements-based PD coefficients, and a stability lobes diagram in orthogonal machining. Initial results reveal that the newly proposed analytical model can eliminate the empirical values identified from time-consuming machining tests and computationally expensive numerical simulations. Next, the introduced method will be extended to PD in three-dimensional (3D) machining processes such as milling with general tool geometries having varying cutting speeds and geometries along the tool axis.

To assess the vulnerability of deep learning in the physical world, recent works introduce adversarial patches and apply them on different tasks. In this paper, we propose another kind of adversarial patch: the Meaningful Adversarial Sticker, a physically feasible and stealthy attack method by using real stickers existing in our life. Unlike the previous adversarial patches by designing perturbations, our method manipulates the sticker's pasting position and rotation angle on the objects to perform physical attacks. Because the position and rotation angle are less affected by the printing loss and color distortion, adversarial stickers can keep good attacking performance in the physical world. Besides, to make adversarial stickers more practical in real scenes, we conduct attacks in the black-box setting with the limited information rather than the white-box setting with all the details of threat models. To effectively solve for the sticker's parameters, we design the Region based Heuristic Differential Evolution Algorithm, which utilizes the new-found regional aggregation of effective solutions and the adaptive adjustment strategy of the evaluation criteria. Our method is comprehensively verified in the face recognition and then extended to the image retrieval and traffic sign recognition. Extensive experiments show the proposed method is effective and efficient in complex physical conditions and has a good generalization for different tasks.

The research-based installation, The Rise, is led by the concept of a growing architecture able to sense and dynamically adapt to its environment as it grows into form while continuously reacting to its own material performance and behavioural constraints. This process is enabled through the careful integration of digital simulation techniques with multi-hierarchical generative design approaches. Aggregations of variably sized bundles of rattan core multiply, bend, branch and recombine into a distributed assembly that manifests an alternative to traditional structural systems. The hybrid approach links a material system with simulation and the iterative generation of geometry through a process of calibration at different stages of design. The project leverages emerging computational strategies for growth in a model for an architectural practice that engages the complexity and interdependencies that characterise a contemporary design practice.

Effects of two meso-scale geometry generation approaches on finite element predictions effective elastic properties of an orthogonal 3D woven composite are studied in this paper. In particular, one model is created by simulating the weaving process in the software DFMA (Kansas State University). The second model is created by directly processing X-ray microtomography (μCT) data. Experimental measurements of transverse Young’s moduli are used to inform the accuracy of the predicted elastic results. In both cases, a unit cell with in-plane periodic boundary conditions is modeled, which has not been previously done in the case of μCT-based models. The effect of high frequency oscillations in tow element orientations imparted by a wavy centerline (artifact of μCT image processing) on the elastic properties is studied. The μCTbased model is then used to simulate tension-to-failure.