Results of the FY2023 MITOU Advanced Program

In-page Links

Results of the FY2023 MITOU Advanced Program

PM: ISHIGURO Hiroshi

Development of a digital twin of human traffic that is automatically constructed from camera images

This project is designed to implement a digital twin for analyzing and predicting the flow of people in public spaces and commercial facilities. In this project, we will develop a unique technology that uses camera images to learn the characteristics and movement patterns of the target space and automatically construct a digital twin. This will make it easier to predict the flow of people and verify measures, and will make it possible to test scenarios such as crowd control and advertising placement in a virtual space. In addition, it reduces the burden of the necessary work each time the target space is set, and it also enables the sharing and collaboration of learning results in different spaces. This project is expected to lead to the widespread use of digital twins of human flow and the realization of safe and comfortable urban spaces.

• View the details Development of a digital twin of human traffic that is automatically constructed from camera images (in Japanese)

Creators: YASUDA Shohei, FUMIYAMA sou, KATAYAMA Hiroki, YAMANISHI Hiromasa

Ubiquitous Force Sensing Platform for Social Deployment

This project aims to develop a platform that uses ubiquitous force sensing technology to turn an entire room space into a omnipresent touch sensor. The platform will leverage distributed force sensors with dedicated algorithms to enable interior surfaces such as desks, shelves, and floors to be used as interfaces, providing an innovative spatial experience. The platform provides a developer toolkit, hardware, and software to help engineers and designers more easily prototype and create rich spatial experiences. Through the seamless integration and physical and virtual spaces, we also aim to improve the value of spatial experiences for all the space owners and designers.

• View the details of Ubiquitous Force Sensing Platform for Social Deployment (in Japanese)

Creators: YOSHIDA Takatoshi, KONDO Toyoki, WATANABE Takafumi, HASHIMOTO Daiki

PM: URUSHIBARA Shigeru

Merchandise Exchange App: Revolutionizing Fandom Life

This project aims to develop “Treportal,” a specialized merchandise exchange platform for dedicated fans. The platform is designed to meet the needs of enthusiasts and streamline the exchange of goods, moving beyond the limitations of traditional social networking services. Treportal offers three core features: posting exchange offers, searching for items, and messaging. These features enable users to easily list items, find what they’re looking for, and initiate exchanges. The platform’s structured exchange process is designed to facilitate smooth transactions without the need for direct messaging.
We will develop both a web application and a smartphone app, with ongoing improvements based on user feedback. To attract and engage users, we plan to implement marketing strategies such as CD releases and giveaways, aiming to grow our user base and increase successful exchanges. Additionally, we will establish a company and develop an intellectual property strategy to promote and monetize the service. This project is expected to invigorate the goods exchange market and contribute to the growth of the content industry.

• View the details of Merchandise Exchange App:Revolutionizing Fandom Life (in Japanese)

Creators: KAMIYA Yuri, HAYASHI Yuri, KAMIJO Yurina, UEYAMA Yukino, NAKAJIMA Mayu, HONDA Shimon

PM: SHUDO Kazuyuki

AI management system that effectively utilizes data assets in the animation production process

This project aims to improve efficiency and communication in anime production. Anime production is a vital industry that enhances Japan's soft power, but the evolution of digital technology has made the production process more complex and increased the burden on creators. AI Mage is designed to streamline communication in anime production by providing a feature that can instantly present reference images. This is expected to accelerate the production process, improve quality, and reduce production costs. AI Mage will create a database of intermediate deliverables in anime production and utilize AI technologies for functions like keyframe extraction and character expression cropping, offering efficient search capabilities. The implementation of AI Mage is anticipated to enhance both the efficiency and quality of anime production, bringing innovative changes to the entire anime industry.

• View the details of AI management system that effectively utilizes data assets in the animation production process (in Japanese)

Creators: XIN Zhang, SU ZiXiong

Development of a fatigue estimation and reduction system using HMD

This project aims to develop technology that utilizes the characteristics of HMD (head mounted display) to monitor the health of users and contribute to health promotion. Specifically, we will develop a model for predicting eyestrain and a function for stretching the eyes using eye tracking, and provide a VR work application for predicting and reducing fatigue. We will also develop an application for analyzing the focusing function of the eyes, and provide these technologies as an SDK that can be used with other VR devices. The novelty and superiority of this project is that it has made it possible to quantitatively measure eye strain and eye adjustment functions, which were previously difficult to do, and to develop methods for reducing eye strain that make use of the visual presentation characteristics of HMDs. In order to spread this technology, we will conduct demonstration experiments at VR game companies and work on medical device applications, and we expect that it will contribute to improving the quality of health management and training in workplaces and educational fields.

• View the details of Development of a fatigue estimation and reduction system using HMD (in Japanese)

Creators: MOMIYAMA Haruki, IKUTA Mitsuki

PM: HARADA Tatsuya

Development of a feedback device to improve the quality of chest compressions

This project aims to develop a device that visualizes and quantifies the quality of chest compressions and provides feedback to rescuers, with the goal of improving the survival rate of patients experiencing out-of-hospital cardiac arrest and preventing hypoxic encephalopathy. This device is attached to the cardiac arrest patient and quantifies the quality of chest compressions by estimating the internal carotid artery pressure. It then displays this information on a screen, providing visual feedback to rescuers. The device is disposable and can be easily attached without interfering with the flow of life-saving procedures. It also enables monitoring of circulatory dynamics data from the onset of cardiac arrest until hospital transport. It is expected that this device will reduce the mortality rate of out-of-hospital cardiac arrest patients by improving the quality of chest compressions and help prevent hypoxic encephalopathy. In the future, it may also have potential applications in the diagnosis and monitoring of other diseases.

• View the details of Development of a feedback device to improve the quality of chest compressions (in Japanese)

Creators: EKUNI shota, MISAWA Toshihide, TANAKA Yuki, KIRIYAMA Hiroyuki

Development of an Automated Fetal Ultrasound System: Advancing Multi-Angle Echo Technology

Monitoring fetal health is essential for safe pregnancy and delivery. Although fetal ultrasound is a valuable, non-invasive tool, it requires expertise and manual scanning, which limits its accessibility. To address these challenges, we develop a new multi-angle ultrasound system, featuring multiple probes placed around the abdomen to enable automated, quantitative, and 3D imaging of the fetus. This system allows pregnant women to conduct self-scans, providing real-time and long-term fetal health evaluations, including estimated weight, amniotic fluid volume, and fetal movements.

The innovation lies in the probe arrangement, which captures wide-ranging ultrasound waves to create detailed 3D reconstructions, reducing reliance on highly specialized equipment and physicians. Automating fetal ultrasonography allows for more frequent screenings, which can help detect abnormalities at an earlier stage and improve perinatal outcomes. This approach addresses the global shortage of obstetricians and could prevent up to half of the 4 million annual perinatal deaths and cerebral palsy cases worldwide.

• View the details of Development of an Automated Fetal Ultrasound System: Advancing Multi-Angle Echo Technology (in Japanese)

Creator: OGASAWARA Jun

Development of Adaptive Learning System in Sports

The Adaptive Learning System in Sports project aims to develop an application that provides the best practice methods tailored to the individual in order to improve their sports progress more efficiently. This application analyzes an individual's movements, identifies issues, and suggests optimal practice methods, simply by sending videos taken with a smartphone. This shortens the time needed to improve sports and enables efficient practice. Specifically, the application will analyze hitting and propose practice methods for baseball, and based on its success, aims to expand to other sports. The application developed features the use of AI technology to analyze individual movements and suggest optimal practice methods. Compared to conventional services, it also offers comprehensive functions, including highly accurate analysis, practice recommendations, and detailed analysis of equipment and body movements. It is expected that the widespread use of this project will streamline the improvement of sports and become an innovative tool for coaches and sports enthusiasts.

• View the details of Development of Adaptive Learning System in Sports (in Japanese)

Creators: HAYAKAWA Satoshi, FUJINO Rintaro

PM: HIRANO Yutaka

Development of “BLACK STONE BRAIN,” an AGV fleet operation planning optimization system using a combinatorial optimization method

This project aims to promote the introduction of automation technology in the manufacturing and logistics industries in order to respond to technological innovation and labor shortages. In particular, the aim is to optimize the operational efficiency of automated guided vehicles (AGV). Conventional AGV systems have issues such as long waiting times for arrival, inability to reach maximum speed, and poor scalability. To solve these issues, we have developed a system called “BLACK STONE BRAIN” that uses the status of all AGVs to search for routes. This system optimizes the operation of AGVs, enabling them to operate at maximum speed while avoiding collisions and preventing congestion. It also improves the efficiency of the entire factory by optimizing production schedules. In the PoC conducted during the project period, it was confirmed that the system reduced the operation time by about 1/7 compared to the conventional system. In the future, we aim to popularize this system, focusing on the AGV market.

• View the details of Development of “BLACK STONE BRAIN,” an AGV fleet operation planning optimization system using a combinatorial optimization method (in Japanese)

Creators: INOUE Ryotaro, OSAWA Takuma, TAKATA Yuta, YAMASAKI Taiki

Development of a low-installation-cost, compact AI sorter for recyclable waste that can be adapted to existing facilities

This project aims to develop a low-cost recyclable waste sorter that can be adapted to existing facilities. At intermediate treatment facilities, where manual sorting is the norm, there are issues related to worker skills, safety, and labor costs. However, although existing large-scale sorting machines exist in other countries, they are difficult to install in Japanese facilities. This project aims to develop a compact AI automatic sorter that can be flexibly adapted to existing facilities, thereby reducing labor and improving quality in the sorting process and making Japan a resource-recycling nation. The main body of the sorter and accompanying software were developed, and a demonstration test was conducted. The sorter is designed to be retrofitted to existing facilities and has a unit-type structure to improve processing speed and accuracy. The accompanying software is a system that analyzes refuse with a camera and visualizes the data. The results of demonstration tests have been positive, and the system is expected to contribute to reducing the workload of sorting workers and improving the efficiency of the facility. There is also a lot of demand for quantitative analysis services, and we plan to expand this service to industrial waste as well. This project is expected to have various ripple effects related to resource recycling, such as improving the efficiency of sorting operations and promoting recycling.

• View the details of Development of a low-installation-cost, compact AI sorter for recyclable waste that can be adapted to existing facilities (in Japanese)

Creators: KAMEDA Koki, HOSOYA Tomoki, TOKUNAGA Yuya

Development of an application to optimize the apparel secondary distribution EC purchasing experience with AI

The goal of this project is to develop ReList, an application that uses AI technology to optimize the apparel secondary distribution EC buying experience. The application aims to expand the secondary distribution market and promote the buying and selling of used clothing by generating demand from wearing images and summarizing items that users may want. It is also expected to contribute to the reduction of CO2 emissions.ReList's main function is to list used clothing from partner flea market apps and suggest similar products. In addition, it utilizes real-time recommendation technology and AI generation of wearing images. The app specializes in exploratory purchasing and provides a superior user experience compared to e.g. Google Image Search. Currently, the app is producing results in user traffic, sales activities, and marketing, and is expected to have a ripple effect throughout the secondary apparel distribution industry.

• View the details of Development of an application to optimize the apparel secondary distribution EC purchasing experience with AI (in Japanese)

Creators: KUMAZAWA Ryusei, IKEDA Yuya, ISHII Shuto

PM: FUJII Akihito

Development of Low Latency IP Video Transmission System for Telepresence Technology

This project aims to develop a low latency IP video transmission system. In response to the need for remote control in the construction industry, we are developing a system that can be used for remote control of construction equipment and other remote operations. Specifically, we have developed the “Shunkei VTX” low-latency video transmission devices, the “ShunCar” IP remote control model car, and the “ShunChecker” glass-to-glass delay measurement device. This system solves the problem of video delay in remote control and provides a comfortable remote control experience. This system is expected to play an active role in the construction industry and in areas such as tele-registration robots, contributing to manpower shortages and improved working environments. It is also expected that the widespread use of 5G and 6G will enable video transmission over a wider area with lower latency.

• View the details of Development of Low Latency IP Video Transmission System for Telepresence Technology (in Japanese)

Creator: MIZUNO Fumiaki, YAMAMOTO Kosuke, EBIHARA Yusuke

Realization of Co-Creation with AI in Video Production

Director AI, an AI assistant to assist in video production, was developed with the aim of establishing a new video production workflow by combining the needs of the video industry with technological advances. The project aims to develop an AI assistant to assist in planning and composition tasks, and to utilize AI in the design of video production. Specifically, Director AI is a system that analyzes videos based on user input and responds using the results of its analysis, with the goal of assisting in the production of advertising videos. The novelty and advantage of this project lies in the fact that it utilizes unique analytical data that is different from other services. In the future, the system is expected to be introduced to advertising agencies, video production companies, and other corporate entities to help improve video production efficiency. It is also expected to be expanded to music videos, dramas, movies, and other video production areas.

• View the details of Realization of Co-Creation with AI in Video Production (in Japanese)

Creators: YAMAMOTO Kenta, SUZUKI Ippei

PM: MIKI Hirofumi

Development of Flood Inundation Forecasting Solution

This project aims to develop an application that predicts the extent and depth of flooding at high resolution. Conventional flood forecasting solutions are limited to predicting river levels. Our solution provides local governments and private companies with tools to receive flood inundation prediction which allow them to utilize for more effective decision making regarding disaster prevention actions. Specifically, the project will extend the technology of flood simulation, demonstrate value through damage estimation, and develop real-time flood forecasts. The applications developed in this project will be able to provide high-resolution flood forecasts, which will differentiate them from other services and technologies. The project is also expected to have a ripple effect in the areas of disaster prevention information systems, sensors, and satellite data.

• View the details of Development of Flood Inundation Forecasting Solution (in Japanese)

Creators: KITA Yuki, DEMOTO Satoru

AR Effect Synthesis App for Dancers Using Machine Learning

A smartphone app called “Charmii” uses machine learning to detect dancers' movements and add AR effects. The app aims to expand the dance industry by not only extending the expression of dance and making it easier to convey the appeal of dance, but also by making even those who are not good at dancing look attractive through the power of the effects. In addition, the application is already available to the public and is being used by many users. In the future, we plan to improve the expressive power of the effects and work to spread a new “AR Dance” culture.

• View the details of AR Effect Synthesis App for Dancers Using Machine Learning (in Japanese)

Creators: FUSHIKI Hideki, TATSUMI Keisuke