Toshiba's New SPINEX IoT Architecture Reinforcing Digital Transformation
Toward Realization of Novel Social Infrastructures Utilizing IoT and ICT
Trends in Industrial IoT and Approach of Toshiba Group
Industries throughout the world are entering the Fourth Industrial Revolution, a new industrial era in which the industrial structure itself is being transformed by digitization. Economic values and growth areas have been changing due to the introduction of digital technologies. Accordingly, manufacturing industries have been focusing more on outcome values and experience values of the customer, rather than on functional values of the product as in the past. These trends will have a major impact on the manufacturing sector. In order to meet the needs of this new era, the shift to new business models to gain customer value with digital technologies, such as the Internet of Things (IoT), big data, and artificial intelligence (AI), is becoming increasingly important to enhance competitiveness.
In response to these customer requirements for digital transformation, Toshiba Digital Solutions Corporation has developed a new IoT architecture called SPINEX. Based on the latest technologies, SPINEX incorporates the Toshiba Group's experience and know-how cultivated in a broad range of industrial sectors including the manufacturing, energy, and social infrastructure industries.
New SPINEX IoT Architecture to Support Digital Transformation
NOMURA Shigeo / KISHIHARA Masaki / FUKAZAWA Shigeru
With the recent movement toward digitization applying information and communication technologies (ICTs), various industries worldwide are being compelled to reconsider their business models. In this situation, it is important to transform the conventional business model into a new business model that makes use of state-of-the-art technologies such as the Internet of Things (IoT), big data, and artificial intelligence (AI).
Toshiba Digital Solutions Corporation has developed the SPINEX IoT architecture, which combines the latest technologies with know-how and expertise cultivated through a broad range of achievements in system development, operation, and maintenance in the infrastructure and industrial fields. Our SPINEX IoT architecture makes it possible to not only improve operational efficiency in product manufacturing by linking various devices and products, but also to accelerate digital transformation in order to optimize the customer's business management.
Security Technologies for Industrial IoT Systems
OYA Toshiharu / NAKAMIZO Takanori / MATSUSHITA Tatsuyuki
Industrial Internet of Things (IIoT) systems have recently been attracting attention, particularly in the manufacturing industry. However, since IoT equipment and control systems are connected to information systems in such IIoT systems, the risk of cyberattacks increases.
To enhance the security of IIoT systems, Toshiba Digital Solutions Corporation has developed support technologies for the acquisition of security certifications to secure the equipment and control systems constituting IIoT systems. We have also developed SecuScope, a tool that facilitates the efficient analysis of security threats by taking the characteristics of the target equipment and control systems into consideration based on international security standards. These technologies make it possible to reduce the work required for the acquisition of security certifications for such equipment and control systems and to ensure security for IIoT systems.
Deep Learning Data Augmentation Technique to Improve Image Recognition Accuracy
ITO Hidemasa / IRIMOTO Yuji / KASHIMOTO Shinichi
In recent years, the need has arisen for the support or automatic implementation of inspection and monitoring work applying image recognition technologies powered by artificial intelligence (AI) instead of experts with technical know-how due to the declining labor force accompanying the shift to an aging society with fewer children. Large amounts of actual image data for learning are required in order to improve image recognition accuracy. However, it is difficult to collect a sufficient number of actual images in some types of operations.
Toshiba Digital Solutions Corporation has developed a unique technique to automatically augment image data using generative adversarial networks (GANs), one of the deep learning technologies. We have conducted image recognition simulations using images of power lines to represent inspection and monitoring work, and confirmed that image recognition accuracy is enhanced by learning that incorporates augmented image data when the amount of actual image data is insufficient.
Scene Text Recognition Technology to Obtain Information in Manufacturing Processes
FURUHATA Akio / ONO Soichiro / FUU Shimou
With the accelerating movement toward the certification of traceability and analysis of big data from processing steps in recent years, the acquisition of information on materials directly and indirectly used in manufacturing processes in factories and other production facilities has become increasingly important. Such data are often acquired from barcodes, etc. However, some information, such as lot numbers and expiry dates, is shown only in the form of characters. The acquisition of such text-based data is one of the main problems in this field.
Toshiba Digital Solutions Corporation has been developing a scene text recognition (STR) technology to support the acquisition of information from factory scenes applying optical character recognition (OCR) technology, a field in which the Toshiba Group has accumulated long experience. Since scenes at manufacturing facilities vary greatly according to the situation, we have customized our STR technology to adjust to the environment by selecting and assembling known modules and optimizing their parameters in accordance with the restrictions specific to the site. We have confirmed that our new STR technology has now reached the practical level.
People-Flow Analysis Technologies to Identify Intentions and Situations
MATSUMOTO Nobuyuki / IKUZAWA Takuya / UNE Yasuomi
In recent years, efforts have been increasingly focused on acquiring an understanding of the ambient situation in various commercial buildings and public facilities through analysis of data collected from IoT (Internet of Things) devices.
Toshiba Digital Solutions Corporation has been developing people-flow analysis technologies to identify situations by focusing on people and their movements in sensed data. As part of these efforts, we have developed technologies to estimate congestion levels in buildings as well as to reduce privacy risks in the case of distributing camera images. Moreover, in response to the intensifying need for the detection of suspicious persons and emergency situations, we have also developed a technology to detect distinctive types of behavior, including unusual actions and events, based on either of two approaches according to the use case; namely, a rule-based approach and a machine-learning-based approach. We have confirmed the feasibility of these technologies through field demonstration experiments for the former approach and accuracy verification tests with simulated video images using extras for the latter approach.
Image Analysis Integrating Characteristic – Distinguishing and Learning Technologies to Understand Movement in Closed-Packed Areas
OUCHI Kazushige / KOBAYASHI Daisuke / NAKASU Toshiaki / AOKI Yoshimitsu
Training and tactical analysis utilizing information and communication technology (ICT) has been progressing in the sports world in recent years, and strategy analysis using image recognition has also been attempted. However, as in the case of rugby with 15 players in a team, in a game where the number of players participating is large and contact and close-packed play occurs frequently, game analysis using image recognition is technically difficult and has therefore not been actively utilized.
In response to this situation, Toshiba has developed a hybrid type video analysis system that detects and tracks a ball according to its distinguishing characteristics and detects and tracks players using deep learning technologies with video images from a single camera. Targeted at rugby, one of the sports being specifically promoted in Japan, this system realizes improved detection of players in close-packed areas and allows the trajectory of the ball and movements of the players to be mapped onto a virtual two-dimensional (2D) rugby field. The system can also perform automatic play classification using a deep learning technique, making it possible to save labor in the task of tagging major plays that has conventionally been carried out manually. This system can be applied not only to rugby but also to various other sports, as well as to industrial fields such as work analysis using surveillance cameras.
IoT Gateway Device Realizing Edge Computing for Social Infrastructure and Industrial Fields
NAKAJIMA Hiroshi / MATSUMOTO Kenichiro
In order to accumulate and analyze large volumes of Internet of Things (IoT) device data in the social infrastructure and industrial fields, it is important to achieve a balance between cloud computing resources and edge computing resources. Edge computing enables users to promptly and accurately grasp the actual situation of equipment and fluctuations in its performance using IoT gateway devices. This, in turn, becomes the basis of the more intelligent rich edge computing to create new business value.
Toshiba Digital Solutions Corporation has developed a novel IoT gateway device for edge computing in these fields. This device has an architecture that allows flexible edge computing and ongoing operational improvements by providing software platforms that can configure various devices, data, and data processing operations in supervisory and control systems. Furthermore, we have realized environmentally durable, highly reliable, and securely robust hardware for this IoT gateway device using the comprehensive technologies in these fields accumulated by the Toshiba Group.
Manufacturing Innovations Using IoT and Toshiba Group's Approach
The manufacturing sector is on the threshold of a dramatic transformation through the use of Internet of Things (IoT) data. With imminent changes in the rules of competition that govern manufacturing industries, various approaches have been initiated at both the government and private levels in Japan following the lead of the United States and European countries.
The Toshiba Group has been participating in cross-sectoral activities being carried out by industries, universities, and governments in the U.S., Europe, and Japan related to this trend. Furthermore, Toshiba Digital Solutions Corporation is offering "Meister Series," next-generation manufacturing solutions for manufacturing industries based on a wide range of technology assets of the Toshiba Group, including development experience and know-how in industrial fields. Meister Series provides total support for the three layers of gathering, storing, and utilizing information. We are contributing to innovations in manufacturing industries by reinforcing human skills, which are one of the strong points of Japanese manufacturing, and by supporting the use of IoT at manufacturing sites.