As the digitization of the manufacturing process is accelerating, various data-driven approaches using machine learning are being developed in chemical mechanical polishing (CMP). For a more accurate prediction in contact-based CMP, it is necessary to consider the real-time changing pad surface roughness during polishing. Changes in pad surface roughness result in non-uniformity of the real contact pressure and friction applied to the wafer, which are the main causes of material removal rate variation. In this paper, we predicted the material removal rate based on pressure and surface roughness using a deep neural network (DNN). Reduced peak height (Rpk) and real contact area (RCA) were chosen as the key parameters indicative of the surface roughness of the pad, and 220 data were collected along with the process pressure. The collected data were normalized and separated in a 3 : 1 : 1 ratio to improve the predictive performance of the DNN model. The hyperparameters of the DNN model were optimized through random search techniques and 5 cross-validations. The optimized DNN model predicted the material removal rate with high accuracy in ex-situ CMP. This study is expected to be utilized in data-driven machine learning decision making for cyber-physical CMP systems in the future.

Deep drawing is one of the most crucial processes in sheet metal forming. As for the multi-stage deep drawing process, because of many design parameters it comprises, predicting process results is a difficult and time-consuming task. In this study, to predict process results, the deep neural network was proposed. Seven design parameters were set and their range was determined with references to empirical formulas. Then, we determined prediction outputs, comprising maximum effective strain, minimum thickness, and bottom mean thickness. Five-hundred sampling points were determined using latinhypercube sampling method. According to the sampling points, finite element analysis was conducted to achieve process results. From the data rendered by the finite element analysis, the deep neural network was trained. Then, the deep neural network was tested with an additional 80 test samples to evaluate performance, and its performance was compared with radial basis function kernel support vector regression. The results showed that the relative performance of the deep neural network was superior to support vector regression.

A measurement of a users’ motion is widely attracting attention for a realization of robotic assistance in daily activities. The soft, wearable sensing suit enables the monitoring of outdoor activities, with high wearability and insensitivity to inertial force. In this paper, we propose a novel sensing suit for measuring the multi degree of freedom (multi-DOF) motion of the wrist joints. We used a fabric-based capacitance-type stretch sensor for high adaptivity to a textile form of suits. The sensor was attached to the body link, instead of the wrist joint to reduce the interdependency among each joint axis and the effect of unwanted disturbance. We adopted the Deep Neural Network for calibration, and verified the higher estimation accuracy on the estimation of the multi-DOF wrist motions. The performance validation proceeded with comparing to the linear-based regression, and the root mean-squared error on the angle measurement was improved at slow motion and fast motion. A real-time measurement interface was developed and demonstrated with a frequency of 250 Hz.