In order to proceed development of deep and wide ocean, utilization of underwater equipment is necessary. Especially, recent large task is how to deploy large number of the equipment at the same time, but price of the equipment inhibit realization of the deployment. In this paper, as a method to reduce the price of the equipment, facilitation of the UHF band communication method and its verification is proposed. From number of the experiments, it is indicated that design guideline of propagation channel material, shape, contact place. Also, behavior of the electromagnetic wave in the propagation channel when frequency is changed and propagation length expanding method is indicated.

Many social infrastructures in Japan have been intensively developed since the period of high economic growth, and the aging of them is progressing. Therefore, periodic inspections and repairs are necessary. Inspections require huge costs and take a long time because inspectors visually inspect bridges directly using scaffolds or special cranes. Practical bridge inspection robots are strongly required to reduce the costs and time associated with inspection of bridges. In recent years, various robots have been developed for bridge inspections. However, many of these robots lack the ability to locomote within inspection areas. Therefore, there are many paths which they cannot move under the three-dimensional complex environments of actual bridges. Practical bridge inspection robots are required to be able to transition between a horizontal plane and a vertical plane. Additionally, they are also required to be capable of traveling through bolted splicing parts. However, robots capable of running through these paths have not been reported. We developed a compact wall-climbing robot BIREM-IV which is equipped with four rimless wheels as driving wheels. Strong permanent magnets are attached to its rimless wheels. Therefore, the robot can move around while adsorbing on a steel structure. Furthermore, the running performance of the robot was examined by experiments. It was confirmed that this robot can easily run through bolted splicing parts. Moreover, this robot can surely transit among floor, vertical wall, and ceiling without falling from the structures.

In this paper, we describe a folding ability acquisition of an item of clothes by a dual-arm robot. We focus on how to fold cloth products with placing it on a chest as a folding method that humans often do. This method has the following advantages: (1) As a robot folds cloth products using its body parts, it can finish the work in the space only near the body, while auxiliary tools such as a table is not needed. (2) Since a robot unfolds a cloth product vertically using hands and chest, and then the work is progressed therefrom, it is relatively easy to recognize the shape of the cloth product. To let the robot acquire this folding method, we take ``Teaching by Showing'' approach. Using time-series sensor data obtained by observing a human performer who folds a cloth product, we propose a unified method to obtain folding phases and folding trajectory for a dual-arm robot. We applied time-series data segmentation and reinforcement learning, and confirmed the effectiveness of the proposed method by means of experiments using an actual robot.