No abstract

Research and development of smart vessels has progressed significantly in recent years, and ships have become high-value technology-intensive resources. These ships entail high production costs and long-life cycles. Thus, modernized technical design, professional training, and aggressive maintenance are important factors in the efficient management of ships. With the continuing digital revolution, the industrial shipbuilding applicability of augmented reality (AR) and virtual reality (VR) technologies as well as related 3D system modeling and processes has increased. However, resolving the differences between AR/VR and real-world models remains burdensome. This problem is particularly evident when mapping various texture characteristics to virtual objects. To mitigate the burden and improve the performance of such technologies, it is necessary to directly define various texture characteristics or to express them using expensive equipment. The use of deep-learning-based CycleGAN, however, has gained attention as a method of learning and automatically mapping real-object textures. Thus, we seek to use CycleGAN to improve the immersive capacities of AR/VR models and to reduce production costs for shipbuilding. However, when applying CycleGAN’s textures to pipe structures, the performance is insufficient for direct application to industrial piping networks. Therefore, this study investigates an improved CycleGAN algorithm that can be specifically applied to the shipbuilding industry by combining a modified object-recognition algorithm with a double normalization method. Thus, we demonstrate that basic knowledge on the production of AR industrial pipe models can be applied to virtual models through machine learning to deliver low-cost and high-quality textures. Our results provide an on-ramp for future CycleGAN studies related to the shipbuilding industry.

The pipe routing of ships has been manually performed by experts, and the design quality depends on the competence of the experts. Therefore, studies on pipe-routing automation and optimization are required. In addition, the pipe-routing task in a ship that requires frequent pipe-routing modifications requires a long time to be optimized. In this study, we developed a methodology that enables a rapid response in situations where frequent pipe-routing modifications are required by applying curriculum learning that can be stably learned by gradually solving easy-to-complex problems. In addition, this study aimed to minimize the length of the pipe and number of bends as an objective function. Finally, the proposed methodology was verified by comparing it with existing studies that used the A*, jump point search (JPS), and reinforcement-learning algorithms to determine the search speed, number of bends, and length of the path.

No abstract

No abstract