In recent years, the growing demand for low volume and high mix production has made it impossible to spend time preparing for production. In the adhesive coating system, it is necessary to generate robot motions by considering the behavior of the adhesive material, but because the behavior is difficult to predict, adjustments are made in kind, and time preparing for production is a problem. In this paper, we propose a method to generate robot motions in the adhesive coating system. A model for predicting the adhesive behavior is considered based on its physical behavior, and the parameters of the model are determined from some actual adhesive data. We also show that it is possible to generate robot motion without in-kind matching by using the model.

Crop diseases are one of the factors contributing to crop yield loss..Because some of these diseases can cause damage to other crops, early detection is necessary. Not only detecting disease strains needs much, but it is also difficult for new farmers to distinguish them; it is necessary to automate the early detection of diseased strains. This paper proposes a disease strain detection method using CAE (convolutional autoencoder). Experimental results show that preprocessing the images and optimizing the CAE structure improved the accuracy.

In the automation of adhesive coating, a problem is that the adhesive is pulled by the robot during coating, which prevents the robot from applying the adhesive according to the target trajectory. To solve this problem, our method uses a linear autoregressive model to learn the relationship between the robot trajectory and the adhesive trajectory. Then, using the adhesive trajectory simulated by the model, the appropriate robot trajectory is determined to match the target adhesive trajectory. The experiment confirms the results of the coating operation on the actual machine to which the proposed method is applied.

In the manufacturing industry, three-dimensional (3D) geometric models are used to represent and analyze cutter-workpiece engagement (CWE). But the dynamic state of the milling process is always represented by the position and shape data of a series of discrete snapshots. Comparing to 3D geometric methods, 4D geometric modeling enables to static and integrated representation of, variable shape, position and orientation of target objects. In other words, 4D models is continuous in time, cutting 4D model into 3D slices can achieve all states at this breakpoint.Authors proposed a strategy for realizing CWE analysis and representation based on 4-dimensional (4D) mesh models. It computes the subtraction between workpiece occupied region (WOR) and tool accumulated volume (TAV). WOR and TAV can represent a still workpiece in spatio-temporal space and the history of tool occupation during the machining process. The basic idea of 4D set operation is to distinguish parts of a model is in or out of another one, then union them to the set operation result. Therefore, the algorithm finds, segments, classifies and combines these parts to construct new meshes. After that, the set operation result enables to be achieved and used to estimate processing surface in CWE analysis.

This study involves the application of EHD phenomena to a pump. EHD phenomena occur when a high voltage is applied to a dielectric liquid, causing the liquid to flow. If the phenomena are used, the liquid's flow can be controlled by arranging electrodes within the liquid and applying a voltage, thus greatly reducing the overall number of parts the pump requires, especially sliding parts, such as impellers. In turn, this makes it possible to reduce the size and weight of the pump. Specifically, the phenomena are created inside the pump by applying high voltage to an electrode installed within a circular tube, which generates a unidirectional flow within the liquid. The purpose of this study, therefore, is to calculate the voltage and velocity distribution by modeling the flow within the EHD micropump and performing a numerical simulation of the flow.

In this research, we propose a method for visualizing the curvature monotonicity regions of planar polynomial Bezier curves using a GPU. The curvature of curves does not change by translation or rotation. Scaling the curves inversely scales the curvature. The curvature monotonicity does not change by translation, rotation, or scaling. Using this property, we propose a method to efficiently visualize the curvature monotonicity regions by introducing the standard form.

In this research, we propose a method for visualizing the curvature monotonicity region for a specific control point of a 3D cubic polynomial Bezier curve using a GPU. A control point is assumed to move in the plane with a constant depth from the viewpoint to the control point, and the curvature monotonicity region is visualized in the plane. Therefore, as the viewpoint position changes, the appearance of the curvature monotonicity region also changes. The proposed method makes it possible to change the shape of a 3D cubic polynomial Bezier curve while checking the region of a control point where the curvature is monotonically varying.

The κ-curve is a recently published interpolating spline which consists of quadratic Bezier segments passing through input points at the loci of local curvature extrema. But their interpolation can only deal with planar curves. Therefore, in this research We propose a method that enables to extend this representation to deal with space curves in a new scheme called κ-space curves.