In Japan, the number of sewer pipes with a useful life of 50 years is expected to increase in the future, and efficient methods are required to maintain them. Various abnormalities occur in sewer pipes. In particular, displaced joints must be inspected and repaired because they are the major factors resulting in road cave-ins. In the current inspection and investigation of sewer pipes, surveyors operate a closed-circuit television (CCTV) camera while checking the monitor on the ground and record the presence of abnormalities and the degree of abnormality in the field or office. However, this method has certain problems in that it is time consuming and expensive to visually evaluate the degree of abnormality and the variation in judgment results among surveyors. In this research, we propose a method for detecting and evaluating displaced joints from the images of sewer pipes using deep learning to address these issues. Then, experiments will be conducted to clarify its usefulness.
In the disaster survey around the water area, there was a difference between the onshore and underwater measurement methods, and it was difficult to immediately grasp the disaster situation. In recent years, research equipment has become unmanned and speedy, and it has become possible to grasp the whole picture immediately. We have developed a method to deliver the situation by simultaneously performing ground survey and bathymetry using ASV equipped with UAV and narrow multi-beam measurement. We report this development process and the results of verification tests.