The Next Generation of Infrastructure Maintenance

Japan faces the dual challenge of aging infrastructure and a declining population, making traditional maintenance methods unsustainable. This presentation explores cutting-edge research and development in AI-driven infrastructure maintenance to address this critical issue.

We begin by showcasing fundamental image analysis AI for tasks like crack detection. Then, we delve into the transformative potential of multimodal technologies, such as vision and language models, and explore the use of generative AI for 3D model reconstruction. These 3D models, including BIM models, FEM models, and virtual tour models, play a crucial role in infrastructure management.

Finally, we will offer a glimpse into the future of infrastructure, highlighting the role of Large Language Models (LLMs) in decision-making and demonstrating the connection between infrastructure management and the future of AI, including the potential of foundation models.