Evaluation of Operability by Different Gesture Input Patterns for Crack Inspection Work Support System

Many concrete structures in Japan, which were built 50 years ago, are facing degradation problems thus the need of regular maintenance is, since cracks are one of the typical deteriorations caused by the aging of concrete structures, its early detection is plays a big role on the maintenance. Many researches have already been conducted on automated crack inspection system. In this study, we develop a crack inspection support system using Mixed Reality Head-Mounted Display (MRHMD) for crack visualization and consider human work evaluation by using hand gesture input. First, we investigated the recognition accuracy of gesture input in HoloLens 2 and proposed the gesture pattern used in the system. Then, we evaluated several gesture patterns to see whether they are suitable for the system or not, and investigated the system performance using hand gesture input. The results show that gestures which are strongly relevant to command have a good performance while gestures that have low relevance shares low preference. We confirmed that for high relevance gestures, gesture input has the potential to improve system performance and efficiency.