Materials Science – Accelerating the Development of New Materials by Introducing New Knowledge Through AI and Big Data Analysis-
AI and big data analysis technologies are expected to be new technologies that will bring innovation to materials informatics by using information processing technologies in the field of materials science. By collecting and using high-quality material data, it is possible to develop unprecedented high-performance materials and improve the efficiency of new material search. In Laboratory of Media Dynamics, we combine traditional materials development techniques (theoretical science, experimental science, and computational science) with AI and big data analysis techniques to bring new knowledge to the field of materials science, which helps increase productivity and create new materials.
Estimation of Material Degradation Area
The rubber used in tires is a composite material made from a combination of polymers, reinforcing agents, cross-linkers, and additives, and its performance is determined by various factors such as the amount of each material and its structure. Therefore, the internal structure of rubber is complex. Manual analysis requires a lot of time and has limitations in terms of analysis accuracy.
In Laboratory of Media Dynamics, we have developed an anomaly detection AI technology that detects structural changes inside rubber. By learning the characteristics of a tire image before use (when it is new), it is possible to detect changes that occur in the tire after use (after wear and aging). The application of this technology reveals the structure of rubber that is resistant to performance degradation and contributes to the development of performance sustaining technology.
Biomimetics Image Retrieval Platform
Biomimetics is a technology that uses the superior functions of living organisms to develop various industrial products, and it is beginning to be used in various industries such as medicine and sports.
In Laboratory of Media Dynamics, we are developing a search engine “Biomimetics Image Retrieval Platform” to support the discovery of superior functions common to living organisms. This retrieval platform makes it easy to find electron microscope images with similar microstructures. The biomimetics image retrieval platform will support the idea of developing new industrial products.
Formulation – Image – Property Relationship Analysis
It is important to understand the relationship between the compounding data used as raw materials for rubber materials, the electron microscope images taken to observe the structure of the synthesized materials, and the physical properties measured from the materials to realize efficient material development.
In Laboratory of Media Dynamics, we have constructed an image generation AI that can generate electron microscope images corresponding to compounding data. In this technology, we succeeded in generating pseudo images similar to actual electron microscope images by learning the relationship between compounding data and electron microscope image corresponding to the compounding data in the image generation model. The realization of this new technology is expected to dramatically accelerate the speed of materials development.
Estimated electron microscope image
Initiatives for Social Implementation
Exhibition at the Tokyo Motor Show
The AI for estimating the degradation area of rubber materials developed by the Laboratory of Media Dynamics in collaboration with Sumitomo Rubber Industries, Ltd. has contributed to the development of flagship tires and was introduced in the chairman’s presentation at the Tokyo Motor Show 2019.
Exhibited at the National Science Museum
We presented an exhibition at the National Museum of Nature and Science to show the results of interdisciplinary collaboration in biomimetics research to the public.
Cooperative Institutions
- Sumitomo Rubber Industries, Ltd.
- Toyota Central R&D Labs., Inc.
- Hitachi, Ltd.
- NEC Solution Innovators，Ltd.