Index

Vol. 81, No. 6, November 2026

Special Reports

Latest Technologies for Railway Infrastructure Digital Transformation Tackling Social Issues

HIRAI Chikara

SUMIYA Akihiko

Digital transformation (DX) through the use of digital technologies, including artificial intelligence (AI), the Internet of Things (IoT), cloud computing, and digital twins, plays an important role in addressing various issues faced by the railway industry, such as critical shortages of maintenance and operational personnel, the need to achieve carbon neutrality, and the need to respond to diversifying customer needs. The movement toward introducing railway DX technologies in various fields, including predictive maintenance, autonomous train operation, train schedule optimization, energy management, and mobility as a service (MaaS), has recently accelerated worldwide.

Toshiba Corporation has been taking the initiative in developing new technologies to address issues related to railway systems, based on its extensive development experience in this field and proprietary cyber-physical system (CPS) technologies. Our approaches to the DX of railway systems are expected to contribute to the optimization of entire railway systems and the realization of sustainable social infrastructure systems.

OHDAKE Tatsuya / HOMMA Akira / SETO Naoto / HATTORI Yohei

In the railway industry, automatic train operations are a focus of growing expectations as a means of addressing the shortage of train operators associated with the declining birthrate and population aging. However, high implementation costs are a critical issue that must be overcome to achieve the practical deployment of autonomous train operation systems.

With the aim of realizing a high-performance, low-cost autonomous train operation system, Toshiba Corporation has developed the following advanced sensor technologies: (1) near- and far-range stereo cameras, (2) compact, high-performance, long-range light detection and ranging (LiDAR), (3) high-accuracy position detection technology, and (4) a train operation assistance system. We are now conducting verification tests of a prototype autonomous train operation system designed to stop trains precisely at station stopping positions without the addition of automated train operation (ATO) equipment for position correction. We are also developing database generation technology required to promote system implementation.

TAMAI Kazuki / OTSUKI Tomoshi / IYAMA Hitoshi

Public transport planning operations still depend on the knowledge and experience of a limited number of skilled workers due to a variety of specific constraints, resulting in considerable time being required for replanning when changes occur. Reducing replanning time has therefore become an issue of vital importance.

To address this issue, Toshiba Corporation is promoting the practical application of optimization artificial intelligence (AI) technologies for TrueLine, a transit scheduling information and communication technology (ICT) solution, to enhance rolling stock and crew scheduling operations.TrueLine makes it possible to automatically create rolling stock and crew schedules through a set of previously separate processes by linking optimization AI technologies via an integrated database. To improve planning quality and speed up rescheduling operations, we have also developed crew duty planning AI in addition to existing optimization AI technologies, including inspection and cleaning planning AI, rolling stock allocation planning AI, and stabling planning AI.

NAMEKI Hideaki / TAKAHASHI Isao

Demand has been increasing among railway companies for improved business efficiency and reduced costs as a result of the shortage of maintenance personnel and the decline in revenue due to depopulation. Calculating the mileage of rolling stock is a critical maintenance task because it is an index for determining the timing of legally prescribed inspections. The mileage is normally calculated based on planned timetables; however, manual correction using actual operational information is required in the event of timetable disruptions, resulting in a significant increase in working time.

Toshiba Corporation has developed a technology to automatically calculate the mileage by estimating actual timetables using both station and global positioning system (GPS) information contained in data collected onboard, and by adding prescribed distances between stations. This technology can reduce the workload while maintaining the same calculation accuracy as conventional methods. We will implement it in remote monitoring systems to support maintenance operations.

SASAKI Tomofumi / MORI Manato / OI Nanoka

The worldwide movement toward the realization of carbon neutrality has recently been accelerating in various industrial spheres. Although railway transportation systems are environmentally friendly, with higher energy efficiency and lower carbon dioxide (CO2) emissions per unit of transport than other transportation systems, railway companies are vigorously implementing various measures for further decarbonization.

In response to this trend, Toshiba Corporation has developed a railway energy management system (EMS) to make effective use of regenerative energy as balancing power in the electricity supply and demand adjustment market through utilization of a traction energy storage system (TESS). A newly developed power demand forecasting technology using smoothing splines is capable of predicting the power received by railway substations during the provision of regenerative energy with higher accuracy compared with conventional technologies. We are also advancing charge and discharge control technologies to increase the volume of balancing capacity provision.

TAKAHASHI Isao / IDE Makoto / MIZUTANI Mami / NISHIKAWA Hiroyuki

As the performance degradation of storage batteries on rolling stock varies depending on the conditions of use, condition-based maintenance (CBM) is required to guarantee the soundness of these onboard batteries. However, it is difficult to directly measure the degradation state of onboard batteries. There is therefore growing demand for battery degradation estimation methods using rolling stock data.

Toshiba Corporation has developed a cloud-based onboard battery degradation diagnosis system using data from Internet of Things (IoT) devices on rolling stock. The system is able to reduce data traffic volumes to the cloud service to approximately 7% of the total data volume by extracting only data required for degradation diagnosis, and to diagnose the degradation status of onboard batteries by estimating the capacity and internal resistance from these data. The results of verification tests using actual data have confirmed that it achieves an estimation error within approximately ±5%, thereby making it possible to determine the appropriate timing for replacement of onboard batteries.

HANASHIMA Yumi / INUKAI Yuki

Maintaining regional public transportation services is becoming increasingly difficult in Japan, especially in rural areas, due to population decline and shortages of drivers. Local governments are undertaking initiatives to reconstruct such services.

To support the successful reconstruction and sustainable operation of regional public transportation services, Toshiba Corporation has developed a digital platform concept to provide end-to-end support from planning through implementation to evaluation and enable continuous improvement cycles by integrating regional and public transportation data and linking data from the planning, implementation, and evaluation processes.

YANAGISAWA Ryuzo / OHYAMA Karin / HAKOZAKI Tomonori

In line with recent trends in the digitalization of tickets and the elimination of magnetic tickets in the railway sector, attention has been focused on QR code-based digital tickets (hereafter referred to as “QR tickets”). There is also a need to enhance collaborative services with external service providers via QR tickets.

Against this background, Toshiba Corporation has developed Doco-Tick, a service combining a QR ticketing system with an open-source service for marketing applications of QR tickets. Doco-Tick achieves lower equipment costs and higher maintainability compared with conventional magnetic tickets. Through proof-of-concept (PoC) studies using QR tickets in cooperation with railway operators and other service providers, we have confirmed that Doco-Tick can visualize the movement patterns of users and can be effectively utilized for marketing.

Feature Articles

KOMODA Kazuma

In the logistics and physical distribution field, there is a growing need for technologies that can stably recognize a wide variety of objects to automate picking operations in physical distribution warehouses, due to the expansion of e-commerce and workforce shortages. However, conventional deep learning-based recognition models not only require a large number of labeled training images for each new object, but their recognition performance also tends to degrade in cluttered environments where objects overlap or are partially occluded.

To overcome these issues, Toshiba Corporation has developed a few-shot learning technology that efficiently adds new objects to the recognition model by training on a very small number of labeled images. The technology comprises the following functions: (1) an image selection function based on a diversity score and clustering that takes feature variations in images into consideration, (2) a data augmentation function to obtain images under various conditions, and (3) a knowledge distillation function to maintain recognition performance for previously learned objects. Experimental results have confirmed that the few-shot learning technology can recognize new objects using only one to ten labeled training images without sacrificing recognition performance for previously learned objects.

KOBAYASHI Yoshitake / HAYASHI Kazuhiro / IWAMATSU Nobuhiro

With the expanding use of Linux® in various fields of social infrastructure, there is growing demand for super long-term support (SLTS) of ten years or more to accommodate long product lifecycles and eliminate the need for individual users to track and apply necessary updates after the standard support period ends.

Toshiba Corporation has played a leading technical role in the Civil Infrastructure Platform (CIP) project, which was established in 2016 in collaboration with companies facing common issues, and has promoted activities that contribute to business value creation for participating companies in non-competitive areas, including the development of SLTS kernels, standardization of real-time control, compliance with International Electrotechnical Commission (IEC) security standards, and development of secure update frameworks. We have also developed Skelios, a custom Linux distribution for Toshiba Group’s products that uses common base layers derived from the outcomes of the CIP project, and have applied it to more than 60 product lines. We estimate that Skelios can reduce the software maintenance workload by up to 70%.

TOKURA Ichiro / MIYAMOTO Takahiro / KOZONO Toshiaki

At the Fukushima Daiichi Nuclear Power Station Unit 2, the safe and secure retrieval of fuel from the spent fuel pool (SFP) inside the reactor building, where radiation levels are high, has been a critically important issue.

The Toshiba Group  has developed fuel handling equipment  capable of retrieving fuel through an opening in the south wall of the reactor building. This approach reduces both the dispersal of radioactive dust and disruption to nearby work while limiting the accumulation of water caused by rainwater inflow during demolition of the reactor building. Although the equipment is based on a general-purpose industrial boom crane to shorten the development period, it fulfills the requirements of the nuclear energy sector through enhanced seismic resistance, redundant safety functions, remote-control capabilities, etc. Following successful shop and on-site testing, installation at the site was completed in March 2026.

NAKAMURA Yasuhito / UEDA Hiroyuki / HIDAKA Ryo / NAKAYAMA Jun

Equity investment for asset management has been increasingly encouraged in recent years, exemplified by Japan's policy initiative to establish the country as a leading asset management center. However, periods of sharp equity market fluctuations caused by unexpected factors have occurred repeatedly throughout history, creating a need for technologies to protect assets by suppressing risk, defined as fluctuations in asset value.

In cooperation with Sumitomo Mitsui Banking Corporation (SMBC), Toshiba Corporation has developed two quantum‑driven diversified indices calculated using the Simulated Bifurcation Machine (SBM), equipped with Toshiba’s proprietary algorithm derived from quantum technologies. The two equity indices put into practical use Toshiba’s equity portfolio management strategy using the SBM, which was announced in 2023, by leveraging SMBC’s expertise in financial markets. A key feature of these indices is their ability to reduce risk and enhance resilience during market downturns by using the SBM to solve large-scale combinatorial optimization problems. Backtesting results have confirmed that the indices outperform major conventional indices in terms of investment performance.

Frontiers of Research & Development

Artificial Intelligence (AI)-Based Technology for Efficient Design of Lipid Compositions for Biodegradable Liposomes


*Company, product, and service names appearing in each paper include those that are trademarks or registered trademarks of their respective companies.