Masato Mizukami

Application and performance evaluation of a lifting device with alternating rotation hoist

There exist vertical long ropes that need periodical inspection for infrastructure maintenance. Development of lifting devices along with the ropes is demanded for steady observation at human-unreachable positions. The lifting device based on alternating rotation hoist has been proposed for carrying inspection devices to the target level. This research realizes an automatic movement of the lifting device by computer control for the improvement of speed and accuracy. Furthermore, a new frame which is equipped with the inspection device is developed for saving weight and keeping balance. Through experiments, performances of the new frame and control system are evaluated from the aspects of speed, accuracy, weight and balance. The results show that the newly developed frame has advantages in weight and speed. The inclination of frame and ascender is measured. The error values are calculated to analyze the accuracy. And the developed control system can control frame effectively.

Evaluation of whole circumference image acquisition system for a long cylindrical structures inspection robot

In an inspection of aging infrastructures has become a problem. Vertical long cylindrical structures is one of inspection targets. Therefore, we have been developing an inspection robot system for these structures. In order to inspect a surface condition of these structures, whole circumference images need to be acquired. In this peper, we implemented a whole circumference image acquisition system for an inspection of these structures in the inspection robot. This system is based on proposed multi-view imaging theory using two cameras and three mirrors. We also evaluated whole the circumference images which is obtained by the developed system using a monochrome pattern.

Exercise classification using CNN with image frames produced from time-series motion data

Exercise support systems for the elderly have been developed and some were equipped with a motion sensor to evaluate their exercise motion. Normally, it provides three-dimensional time-series data of over 20 joints. In this study, we propose to apply Convolutional Neural Network (CNN) methodology to the motion evaluation. The method converts the motion data of one exercise interval into one gray scale image. From simulation results, the CNN was possible to classify the images into specified motions.