Service using AI to forecast market prices of five vegetables offered free as part of NEDO project - goal is to boost productivity and viability of indoor farms

The New Energy and Industrial Technology Development Organization (NEDO) is working on the Realization of a Smart Society by Applying Artificial Intelligence Technologies project. As part of this, Farmship, Inc. and Toyohashi University of Technology have developed a produce market price forecasting algorithm using AI. They have developed a mechanism of forecasting the weekly market price of five products, including tomatoes and strawberries, in addition to its existing lettuce price forecasting, to a high degree of accuracy, and launched a free market price forecasting service on March 24.

The production of crops from indoor farms has increased dramatically in recent years, as they can be produced consistently regardless of the weather, unlike those farmed outdoors, and also take up less space. However, the fact that demand for the produce is greatly affected by the price movements in outdoor produce is causing food waste and lost sales opportunities.

To overcome this issue, NEDO has undertaken the Realization of a Smart Society by Applying Artificial Intelligence Technologies project, which involves collecting big data from the process of growing and harvesting produce at indoor farms and transportation, then applying this data to match demand and supply using AI. Based on these, NEDO aims to cut food waste at indoor farms and reduce lost sales opportunities by streamlining the entire value chain, such as by optimizing every process, from managing the growth of crops to transportation.

In the project, Farmship, the University of Tokyo, and Farmship subcontractors Toyohashi University of Technology and Pi Material Design Inc. researched and developed a demand forecasting system using on-site data collected since FY2018 at each stage from production to transport and sales, as well as a growth management system to feed the forecast findings back to farms. Since November 2019, a trial has been underway forecasting the price of lettuce on a monthly basis.

As part of the trial, the University of Tokyo developed sensor technology to improve lettuce cropping efficiency, Toyohashi University of Technology worked on demand forecasting technology, Pi Material Design developed crop growth forecasting technology and Farmship designed technology to improve the efficiency of the whole system.

In addition, NEDO, Farmship and Toyohashi University of Technology kept proceeding with developing a vegetable market price forecasting algorithm using AI. By upgrading the basis of supply and demand matching for vegetable at indoor farms, they were able to improve the precision of the monthly forecasting algorithm used to date at Ota Wholesale Market, Central Wholesale Market, Tokyo, and began offering a free service that provided a weekly forecast from March 24. The categories have also extended beyond lettuce to include five product lines including the heavily-produced and distributed tomatoes and strawberries, which can currently be forecast with a high degree of precision.

This initiative will allow more precise demand forecasting of the five products, and is expected to bring the benefits of reduced waste and loss of sales opportunities.

Farmship was awarded the FY2020 Sixth JEITA (Japan Electronics and Information Technology Industries Association) Venture Award on March 24 for the market price forecasting algorithm, which was praised as an initiative that can be harnessed in indoor farming technology.