ABEJA Singapore Pte. Ltd. Launches Full-scale Cooperation with Denso International Asia Co., Ltd to Improve Work Efficiency in Thai Factory
– Rationalizing Work Analysis Utilizing AI to Optimize Operations on the Factory Floor –
ABEJA Singapore Pte. Ltd., the local subsidiary of ABEJA, Inc. (head office: Minato, Tokyo; Representative Director: Mr. Yousuke Okada; hereafter, “ABEJA”), and Denso International Asia, Co., Ltd (DIAT), which is a member company of the Denso Corporation group (Kariya, Aichi Prefecture; President & CEO: Mr. Koji Arima) located in the Kingdom of Thailand (hereafter, “Thailand”) and serving as the company’s headquarters in the Australia-Asian region, began a collaboration designed to improve work efficiency in DIAT’s plants on November 1st, 2018. It can also be revealed that, in advance of this collaboration, the two companies conducted a joint demonstration pilot for some five months, from April to August 2018.
Since ABEJA Singapore began business in April 2017, it has put its core technology, “ABEJA Platform,” to good use in leveraging the AI social implementation know-how (including deep learning) that it has developed within Japan, working collaboratively with local companies, including academic institutions and the healthcare industry, in the Republic of Singapore and other ASEAN nations.
DIAT was established in February 2007 in the suburbs of Bangkok, Thailand, as the Denso group’s Australia-Asian regional headquarters, thereafter engaging in the formulation of key policies and business plans, as well as management and support services for production planning, information systems, human resources development, and more. DIAT’s Technical Center develops products optimized for the Thai domestic market, driving improvements in the Denso group’s competitiveness in the Australia-Asian region.
The aim of this collaboration with ABEJA is to use AI technologies, including deep learning, to automate the measurement and analysis that DIAT has been carrying out for each manufacturing process as part of its drive to increase work efficiency on the factory floor. DIAT also intends to use the analytical findings obtained from this collaboration to reveal differences in operational speeds and operational procedures, thereby making manufacturing operations visible in order to optimize overall factory operations.
In the joint demonstration pilot conducted from April 2018, deep learning was used to calculate variation in worker posture using data from work footage taken in a Denso Thailand plant in Eastern Thailand, and to define the work “cycle” by taking the points in time when the workers’ postures match as the beginnings of new work tasks. This not only allows the work time per cycle to be analyzed but also for specific tasks to be analyzed from work footage, using postural continuity, in addition to conventional posture recognition, to calculate the time per specific task within the cycle and thereby make precise estimates possible. As a result of this joint demonstration pilot of approximately five months, the team were able to build a trained model with a degree of accuracy similar to that of work analysis performed visually by a human. In September 2018, ABEJA began supplying a “video annotation” tool for labelling specific scenes in video data and employing it in the process of creating trainer data from the video footage, with both companies working to build a highly accurate trained model.
DIAT and ABEJA Singapore aim to further improve the accuracy of the trained model built during the joint demonstration pilot, in preparation for its actual operation in the plant, as well as to expand the applicable scope of the techniques established during the joint demonstration pilot, deploying them first across production sites in Denso group-operated plants throughout Asia, and then globally. The companies will also strive to bring to factory floors innovations designed to play a role in solving sharp rises in labor costs, which have been an issue for expanding business in ASEAN in recent years.
Comments from both Companies on this Collaboration
Mr. Masao Suematsu, COO & Executive Vice President, Denso International Asia Pte., Ltd. (DIAS)
The objective of this collaboration is to further pursue the lean manufacturing capabilities that Denso has been assiduously developing. We have embarked, firstly, on making the task of operational analysis more efficient, and aim, in future, to use AI to identify problem areas within production sites that have not been previously visible, in order to achieve an even higher level of production management.
Mr. Naoki Tonogi, Managing Director, ABEJA Singapore Pte., Ltd.
I am delighted that the demonstration pilot, designed to contribute to ASEAN’s future and begun in April this year with the ground-breaking manufacturing company Denso International Asia (which enjoys a top market share globally), has successfully progressed to full-scale development. Together with Denso International Asia, ABEJA hopes to put each company’s areas of expertise to good use and to showcase to the world how man and machine can work in harmony in manufacturing in response to social challenges in ASEAN.
■ About Denso International Asia Co., Ltd. (DIAT)
Company name: Denso International Asia Co., Ltd. (DIAT) Established: February 2007 Location: 888 Moo 1, Bangna-Trad Km. 27.5, Tambol Bangbo, Amphur Bangbo, Samutprakarn 10560 Equity interest: Denso International Asia Pte., Ltd. (DIAS) 100% Representative: Mr. Shinsuke Yamaguchi, President Areas of business: Regional headquarters for the Asia region, and design and development of auto parts
■ About ABEJA Singapore
Company name: ABEJA Singapore Pte. Ltd.
Established: March 2017 Location: 36 Robinson Road, #02-01 City House, Singapore 068877 Equity interest: ABEJA, Inc. 100% Representative: Mr. Naoki Tonogi, Managing Director Areas of business: Expanding “ABEJA Platform” and “ABEJA Insight for Retail” business, and developing new lines of business
■ About ABEJA, Inc.
ABEJA is a venture company that uses deep learning, which automatically extracts, without human intervention, features that appropriately represent the data from accumulated big data, in order to drive innovation in business in a variety of scenarios across a broad range of industries. Since it was founded in 2012, ABEJA has worked in cooperation with university faculties in Japan that specialize in AI-related fields, conducting research into technologies such as deep learning and developing, in-house, all of the deep learning technologies used in the services that it provides. Its core technology, “ABEJA Platform,” is used in collaborations with companies such as Daikin Industries, Ltd.; Musashi Seimitsu Industry Co., Ltd., providing a variety of solutions to a broad range of industries and achieving an installed base of AI systems operational in over 100 companies. ABEJA also provides “ABEJA Insight,” a package service for the retail and distribution industry, the manufacturing industry, and the infrastructure industry, with a track record of deployment in approximately 100 companies across more than 520 outlets (as of end of Nov 2018), mainly in major Japanese retail and distribution companies, including Parco and ICI Ishii-Sports, Inc. In March 2017, ABEJA established a local subsidiary in the Republic of Singapore and is working to develop business in ASEAN nations and expand globally.
Company name: ABEJA, Inc.
Representative: Mr. Yousuke Okada, Representative Director
Location: 10th Floor, NBF Platinum Tower, 1-17-3 Shirokane, Minato-ku, Tokyo
Areas of business: AI business solutions using deep learning for social implementations