Few Samples of SAR Automatic Target Recognition Based on Enhanced-Shape CNN

Synthetic Aperture Radar (SAR), as one of the important and significant methods for obtaining target characteristics in the field of remote sensing, has been applied to many fields including intelligence search, topographic surveying, mapping, and geological survey. In SAR field, the SAR automatic target recognition (SAR ATR) is a significant issue. However, on the other hand, it also has high application value. The development of deep learning has enabled it to be applied to SAR ATR. Some researchers point out that existing convolutional neural network (CNN) paid more attention to texture information, which is often not as good as shape information. Wherefore, this study designs the enhanced-shape CNN, which enhances the target shape at the input. Further, it uses an improved attention module, so that the network can highlight target shape in SAR images. Aiming at the problem of the small scale of the existing SAR data set, a small sample experiment is conducted. Enhanced-shape CNN achieved a recognition rate of 99.29% when trained on the full training set, while it is 89.93% on the one-eighth training data set.

Determination of optimal shot peen forming patterns using the theory of non-Euclidean plates

We show how a theoretical framework developed for modelling nonuniform growth can model the shot peen forming process. Shot peen forming consists in bombarding a metal panel with multiple millimeter-sized shot, that induce local bending of the panel. When applied to different areas of the panel, peen forming generates compound curvature profiles starting from a flat state. We present a theoretical approach and its practical realization for simulating peen forming numerically. To achieve this, we represent the panel undergoing peen forming as a bilayer plate, and we apply a geometry-based theory of non-Euclidean plates to describe its reconfiguration. Our programming code based on this approach solves two types of problems: it simulates the effect of a predefined treatment (the forward problem) and it finds the optimal treatment to achieve a predefined target shape (the inverse problem). Both problems admit using multiple peening regimes simultaneously. The algorithm was tested numerically on 200 randomly generated test cases.

Design of Stackable Origami Structures With Elastic Deployment Capabilities

This work presents a flexible type of origami structure that may be elastically deployed from a compact stacked form to a freeform target shape. The design process enables a target surface mesh to be converted into a compact stacked structure that may be deployed through the release of elastic energy stored in the folds. The process begins by finding a non-branching path passing once through each face in the target mesh. The edges of the target mesh not included in the path are cut and elastic smooth folds are introduced along those crossed by the path. The introduced smooth folds are folded in a sequence of ±180° along the path to create a stack. The structure transforms from the stacked form towards the target shape through the release of the stored elastic energy generated during stacking. The design framework considers the strain energy needed to sustain transformation and the required sizing of the smooth folds. The resemblance of the designed target shape with smooth folds compared to the target mesh is studied, and the significant volume saving when the structure is stowed in the stacked form is quantified. Examples showing the application of the design process to a diverse set of target meshes are provided. Proof-of-concept prototype fabrication using a 3D printer demonstrates the feasibility of the design approach. The results reflect the benefits of deployable stacked origami structures and show volumetric space savings from 50% to 90% while preserving around 80% of the target mesh area after the elastic smooth folds are introduced.

The influence of multi-point roller-dies on the shape deviation differences of profile in flexible stretch-bending forming

In the process of flexible 3D stretch bending, the shape deviation difference between the contact zone and non-contact zone is studied. It is obvious that in the contact zone, the die regulates the deformation of the profile to make it conform to the target shape with small shape deviation; in the non-contact zone, the profile has no die restriction and deviates from the target shape with large shape deviation. When the dies are placed equidistantly along the x-axis, the shape deviation of the non-contact zone near the clamp side is greater than that near the middle of the profile. Arrange the distance between adjacent dies in equal ratio along the x-axis, so that the spacing near the clamp side is a little smaller, and the spacing near the middle of the profile is a bit larger. The difference between the shape deviation of the non-contact zone profile near the clamp side and the middle of the profile decreases, and the maximum shape deviation is reduced, which greatly improves the processing accuracy and quality. However, with the increase of the distance difference between adjacent dies, the shape deviation difference of the non-contact zone near the middle of the profile also increases greatly. Although the clamp side decreases, the maximum shape deviation has become the shape deviation of the profile in the non-contact zone near the middle of the profile.

Generalization of motor learning in psychological space

ABSTRACTTo generate adaptive movements, we must generalize what we have previously learned to novel situations. The generalization of learned movements has typically been framed as a consequence of neural tuning functions that overlap for similar movement kinematics. However, as is true in many domains of human behavior, situations that require generalization can also be framed as inference problems. Here, we attempt to broaden the scope of theories about motor generalization, hypothesizing that part of the typical motor generalization function can be characterized as a consequence of top-down decisions about different movement contexts. We tested this proposal by having participants make explicit similarity ratings over traditional contextual dimensions (movement directions) and abstract contextual dimensions (target shape), and perform a visuomotor adaptation generalization task where trials varied over those dimensions. We found support for our predictions across five experiments, which revealed a tight link between subjective similarity and motor generalization. Our findings suggest that the generalization of learned motor behaviors is influenced by both low-level kinematic features and high-level inferences.

Embedded Actuation for Shape-Adaptive Origami

Origami-inspired approaches to deployable or morphing structures have received significant interest. For such applications, the shape of the origami structure must be actively controlled. We propose a distributed network of embedded actuators which open/close individual folds and present a methodology for selecting the positions of these actuators. The deformed shape of the origami structure is tracked throughout its actuation using local curvatures derived from discrete differential geometry. A Genetic Algorithm (GA) is used to select an actuation configuration, which minimizes the number of actuators or input energy required to achieve a target shape. The methodology is applied to both a deployed and twisted Miura-ori sheet. The results show that designing a rigidly foldable pattern to achieve shape-adaptivity does not always minimize the number of actuators or input energy required to reach the target geometry.

An Efficient Ultraprecision Machining System Automating Setting Operations of Roughly Machined Workpiece

Ultraprecision machining is required in many advanced fields. To create precise parts for realizing their high performance, the whole machining process is usually conducted on the same ultraprecision machine tool to avoid setting errors by reducing setting operations. However, feed rate is relatively slow and machining efficiency is not so high compared to ordinary machine tools. Thus, the study aims to develop an efficient ultraprecision machining system including an industrial robot to avoid manual setting and to automate the setting operations. In this system, ultraprecision machining is conducted for the workpiece having a shape near the target shape, which is beforehand prepared by ordinary machine tools and is located on the machine table by means of an industrial robot. Since the setting errors of the roughly machined workpiece deteriorate machining accuracy, the differences from the ideal position and attitude are detected with a contact type of on-machine measurement device. Numerical control (NC) data is finally modified to compensate the identified workpiece setting errors to machine the target shape on an ultraprecision machine tool. From the experimental results, it is confirmed that the proposed system has the possibility to reduce time required in ultraprecision machining to create precise parts with high efficiency.