Theme 7
SDGs and Sumitomo: New Work Styles

Agriculture in Japan faces grave challenges due to the shrinking population working in the agricultural sector and gradual abandonment of farmland. The Japan Research Institute (JRI) is seeking ways to transform these challenges into opportunities to make farming a high-value-added, profitable business sector.

Called “DONKEY,” JRI’s autonomous multifunctional robot is designed to support farmers in labor-intensive tasks such as harvesting, transportation and weeding. JRI engaged in joint research with Keio University on this robot. In November 2017, JRI established a development consortium with Keio University, various companies in the equipment manufacturing, finance and trading sectors, and the town of Motegi in Tochigi Prefecture, where the mountainous terrain makes farming particularly challenging.

The agricultural robots already in use tend to be costly yet offer limited functionality. A robot specialized to harvest a certain crop goes into barn storage once the harvest is finished, making it an expensive outlay for farmers cultivating several different crops.

In terms of its concept, DONKEY, consisting of a base module plus various interchangeable attachments, emphasizes multifunctionality. DONKEY can handle a range of tasks, including weeding, sowing, transporting or harvesting various crops; countermeasures to protect crops from wildlife; and preparing or managing soil. Its design means it can be used for a much greater portion of the year, increasing the cost benefit. DONKEY is designed to be the farmer’s right-hand man.

If, as envisaged, not only local farmers but also local workshops, small and medium-sized enterprises, students and other players become involved in the development and fabrication of attachments and expansion modules for DONKEY, the multifunctional robot is expected to spur regional revitalization extending beyond the agricultural sector.

A further benefit of the robot is that it collects farming data, which could enable farmers eventually to make more accurate harvest predictions and raise productivity while also facilitating the transfer of know-how to the next generation.

The development consortium finished its work at the end of March 2018. Field-testing started in Motegi in April 2018 and DONKEY has been further improved, reflecting feedback from farmers.

In view of its potential, DONKEY is an opportunity for JRI to address various social issues and be involved in the revitalization of Japan.

Help desk operations and other customer support services to respond to questions from customers are essential for companies, but human resources for these activities are limited. In many cases, customers do not consult the online FAQs and related answers.

SCSK has developed an AI-based conversational agent called Desse to help address this issue. The system places a character on the company’s website to give automatic answers to questions from customers. Through real-time text-based conversation, Desse can interact with customers to solve issues by engaging in online chat.

A defining characteristic of Desse is that it comprehends natural conversational language, and is equipped with a knowledge database to generate highly relevant answers automatically depending on question content. The technology is an extension of machine translation for automatic translation of text, a field in which SCSK has been conducting research for many years, and is highly reliable.

Numerous companies are already using Desse. It is proving its utility in terms of reducing the need for call center operations, which require deployment of human resources. Introducing Desse allows companies to give customers a greater opportunity to solve the issue without having to phone the call center, which in turn helps reduce the volume of calls. Some clients have reported reductions of almost 50% in customer support call volumes.

Ease of use is another advantage of Desse. For instance, adding assumed Q&A patterns to increase the accuracy of responses would require programming knowledge with conventional approaches. With Desse, all you need to do is enter Q&A text on a spreadsheet software worksheet and upload the relevant file to the server.

A further benefit of Desse is the ability of its Q&A engine to support multiple languages, reflecting its roots in machine translation technology. Desse currently supports seven languages: Japanese, English, Chinese (three languages), Korean, and Thai. Marketing and promotional applications are another possibility via efficient collection of Q&A-generated customer data.

Entrusting some operations to a machine enables the redeployment of freed up human resources to other areas. With its ease of system installation, Desse shows companies one way in which AI could be used in the future.