Service

　Our research is based on the field of “Service Engineering“, which handles “services” from an engineering perspective and discuss methodologies from design to development.

　Most of the research in this group is in the field of “Abnormality Detection / Abnormality Prediction“, which automatically detects or predicts the “Abnormal products/behaviours” that causes problems and disadvantages to the industry. “Abnormal products/behaviours” are usually rarely observed, such as failure of machinery and equipment rarely occurring in factories and defective products rarely generated during processing of products. Our research is now being conducted in cooperation with various companies and research institutions.

　There are various methods for abnormality detection and abnormality prediction, but they are roughly summarized as the figure shown below.
　 First, we install sensors, microphones, cameras, etc. to machines/equipment, which is the target of abnormality detection/prediction, and collect data regularly and periodically. 　 Second, we develop an anomaly detection/prediction model based on machine learning (deep learning). After modelling, we test whether it is possible to detect/predict abnormal products/behaviours that are actually occurred.
　Some models and research results developed by this research group are patented and some are used in actual worksites.

　As mentioned before, there are various ways for “Anomaly Detection/Prediction” such as defective product detection, failure prediction, and soundness monitoring. It is important to choose carefully the type of data to be acquired and machine learning/deep learning methods to be used as detection/prediction model because these are depending on the target of detection/prediction and the type of abnormal. In addition, by clarifying what problems are present, in what process/order and to what extent to be solved, we will approach the realization of clear and robust abnormality detection and abnormality prediction. With careful consideration of the above points, our group is engaged in research.

　We have focused on anomaly detection and anomaly prediction, but this research group is also engaged in various other researches based on service engineering, such as estimating store wait time, estimating work efficiency, and so on.