Vol. 77, No. 5, September 2022

Special Reports

Highway and Traffic Solutions Underpinning Mobility in Era of CASE

OGUCHI Takashi

KOBAYASHI Hiroshi / UENO Hideki

Innovation activities related to CASE (connected, automated, shared and service, electric) mobility have recently become a focus of attention as a concept of next-generation mobility services that are expected to bring about a once-in-a-century transformation of the automobile industry. This transformation will also necessitate a major evolution of the road systems that support the vehicle traveling environment. From the perspective of public systems, on the other hand, road systems must be able to provide services to ensure a safe and comfortable driving environment for all traffic including general vehicles while taking CASE into consideration.

In response to this trend, the Toshiba Group has been engaged in research and development aimed at realizing road systems with higher reliability through the provision of highway and traffic solutions corresponding to CASE.

SHIMOKAWA Yusuke / OHBA Yoshikazu

Rear-end collisions with vehicles that have stopped at times of traffic congestion are one of the main causes of traffic accidents on expressways. An effective means of reducing such accidents is to provide drivers with prediction information when there is a high probability of an accident occurring.

Toshiba Infrastructure Systems & Solutions Corporation is aiming at the practical application of a method to predict the occurrence of traffic accidents on expressways by means of a self-organizing map, which is a type of neural network, in order to prevent traffic accidents from occurring through the provision of such prediction information to traffic control personnel and drivers. From the results of verification tests using actual data collected from multiple expressway routes, we have confirmed that this traffic accident occurrence prediction method satisfies the target specifications.


Traffic congestion on expressways, which has been a social issue for many years, is often caused by accidents.

In cooperation with Tohoku University, Toshiba Infrastructure Systems & Solutions Corporation is working on the development of a technique utilizing a machine learning method to predict the flow of traffic in the event of traffic accidents on expressways. We have constructed a machine learning model on a trial basis to predict the flow of traffic for 60 minutes after the occurrence of an accident. The results of tests using actual data have confirmed that the traffic-flow prediction model achieves an accuracy rate of 75% or higher as stipulated in the target specifications in more than half of all cases. In the future, we will consider incorporating information related to the scale of traffic accidents in order to further improve the accuracy of the model. We have also conducted studies on the provision of traffic-flow information to drivers based on the prediction results by means of heatmaps of vehicle velocity, making it possible to estimate arrival times and avoid traffic congestion according to the driver’s departure time and present location.


Commercial software has conventionally been used for the operating systems and middleware of many infrastructure systems, including expressway toll collection systems, from the viewpoint of ensuring the stable operation and long-term maintenance of such systems. In recent years, however, there has been a growing need for the replacement of commercial software with open-source software (OSS) in response to the introduction of various new services as well as the changes that taking place in the public administration and business environments. It is necessary to address a variety of issues in this regard, including migration costs, quality assurance, and long-term operation and maintenance.

To resolve these issues, Toshiba Infrastructure Systems & Solutions Corporation has been engaged in the development of a method using migration tools to swiftly and efficiently implement migration from commercial software to OSS for expressway toll collection systems while making use of a system’s existing assets. We have confirmed the effectiveness of this method through trial migrations of applications and data from conventional commercial software to PostgreSQL, an open-source relational database management system, using three types of migration tools.

AKUTSU Yusuke / NUKADA Sunao

The concept of strategic toll rates to optimize traffic flows for more effective utilization of expressways in the Chukyo area (the metropolitan area centering around the city of Nagoya) was announced by the Ministry of Land, Infrastructure and Transport (MLIT) in February 2020. The concept of strategic toll rates to optimize traffic flows for more effective utilization of expressways in the Chukyo area (the metropolitan area centering around the city of Nagoya) was announced by the Ministry of Land, Infrastructure and Transport in February 2020. Subsequently, in May 2021, a new toll rate system was introduced in the Chukyo area in which toll rates are calculated based on the starting and ending points of a journey regardless of the route. In implementing the new system, it was necessary to realize seamless toll rates that eliminated the dependence on individual expressway operators.

Toshiba Infrastructure Systems & Solutions Corporation developed key functions for this project, including the sharing of information on vehicular travel history between expressway operators and the calculation of new toll rates in accordance with various conditions, based on its accumulated experience in the development of the expressway toll collection system delivered to Nagoya Expressway Public Corporation. These functions are contributing to the effective utilization of expressways being promoted by MLIT. 


The collection of data from edge devices including sensors and drive recorders installed in vehicles has become easier in recent years. This has led to the active development of services utilizing edge device data in various business fields, not only the automobile industry but also other related industries such as insurance, security, and towing services.

The Toshiba Group has developed the following functions that can be used to offer various customer support services through the analysis of such data using artificial intelligence (AI) techniques: (1) an analysis function that makes it possible to classify the causes of impacts based on data from edge devices and (2) a reproduction function that makes it possible to reproduce the situation in the event of a traffic accident based on video data from drive recorders.

Feature Articles


In the field services of power plants, demand has been growing for new services in which experienced personnel can simultaneously support personnel working at multiple sites from a remote location so as to improve business efficiency and customer satisfaction. However, it has been difficult for camera systems to audiovisually grasp the on-site situation at a level of accuracy comparable to that attained by dispatching personnel to the actual site.

Through a process of collaboration between field service and R&D departments involving actual case studies from a remote location, the Toshiba Group has developed a mobile cloud camera system for field services that has sufficient resolution to solve this problem. We have conducted trials at several power plants in Japan and confirmed that the newly developed mobile cloud camera system can transmit compressed high-definition (HD: 1 280 × 720 pixels) images at a practical data rate of 200 kibibit/s, meeting the requirements for remote support of a wide range of field services.

SHIJO Tetsu / LIN Qiang

The stability of power grids has conventionally been maintained by the large inertia of turbines and other rotating equipment connected to synchronous generators such as those in thermal and hydroelectric power stations. However, the reduction of inertia due to the rapid introduction of renewable energy in recent years has become a serious issue. In particular, inertia reductions tend to occur during wide-area blackouts caused by disasters, as well as in renewable energy microgrids constructed on isolated islands and in mountainous areas to realize self-sufficiency in electricity.

In order to develop measures appropriate for such low-inertia power grids, Toshiba has constructed a test environment that can simulate various types of microgrids. This test environment contains five 20 kW battery energy storage systems (BESS), each having a capacity of 14.9 kWh, which can be equipped with either grid-forming (GFM) inverters with inertia-supporting control (supplying virtual inertia) or grid-following (GFL) inverters without inertia-supporting control (without virtual inertia), connected in parallel with a 125 kVA diesel synchronous generator (SG). Through the results of tests, we have confirmed the effectiveness of the operation of five BESS incorporating GFM inverters with inertia-supporting control operating in parallel with a diesel SG for the stabilization of low-inertia microgrids without increasing the load shared by the diesel SG.

OHIRA Hidetaka

There is a growing need for waste treatment facilities in Japan to improve the efficiency of their waste treatment operations, accompanying the rising awareness of environmental issues, tight financial situation faced by local governments, and aging of workers in recent years. In this context, attention is being increasingly focused on the application of automatically controlled cranes using artificial intelligence (AI) technologies to waste pits in order to efficiently stir and transship wastes.

As an addition to its professional service lineup of SATLYS Toshiba Analytics AI services, the Toshiba Group has developed a technique to recognize conditions in waste pits including the type of waste, waste stirring condition, and height of waste, using images captured by a monocular camera. This technique has been incorporated into a fully automatic AI waste crane system developed by Kawasaki Giken Co., Ltd., which makes it possible to precisely and efficiently perform waste stirring and transshipment even in cases where difficult crane control operations are required.

KODERA Shiho / NAKAJIMA Hiroshi / SUGIMOTO Nobuhide

The Toshiba Group has set the goal of becoming a company that delivers new value, contributes solutions to social issues, and helps to achieve a sustainable society through the utilization of cyber-physical system (CPS) technologies.

In CPS, large volumes of data are collected in physical space by means of Internet-of-Things (IoT) technologies, analyzed by digital technologies in cyberspace, and then fed back to physical space to promote advances and improvements. We have now developed the Toshiba IoT Service Factory (TISF) in order to swiftly provide customers with services compliant with the Toshiba IoT Reference Architecture, a common framework to promote the development and operation of CPS. TISF is a development environment that simplifies the contents of CPS services into three types of patterns and maximizes the use of software components to automatically develop new services without the need for conventional system integration (SI), making it possible to rapidly respond to various changes in physical space.

TAKENO Yuishi / NOBUOKA Tetsuya

Self-checkout systems, which provide an environment allowing customers to register purchased products by themselves based on image information captured by a camera on a terminal such as a tablet or smartphone, have become widely disseminated in the retail industry in recent years. However, as price information includes discount labels with different designs of individual stores in addition to the price barcode label of each product, it is essential to reduce the risk of discount label misrecognition so as to avoid mistakes in the amount payable for purchases.

Toshiba Tec Corporation has been developing a discount label recognition function for lightweight applications running on tablets and smartphones. We have now further enhanced the discount label recognition function using an adaptive data augmentation technique instead of the static data augmentation technique conventionally used for training deep neural networks. This technique incorporates the following features: (1) sufficient reduction of misrecognition risk through the application of a dynamic data augmentation algorithm appropriate for discount label designs and (2) adaptive image processing without the use of previous knowledge related to discount label designs. These features make it possible to rapidly accommodate various types of discount label designs at low cost.

Frontiers of Research & Development

Autonomous Mobile Technology to Decrease Intervals between Roll Box Pallets for Efficient Use of Space in Physical Distribution Warehouses

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