Business / Tech

Harnessing the power of AI: Japanese delivery firms and restaurants look to tech to boost businesses

by Osamu Tsukimori

Staff Writer

In the midst of a surge in demand as more people shop online, the parcel delivery sector is struggling to keep up due to a chronic shortage of drivers. Meanwhile, restaurants are struggling to find ways to reduce waste in an industry notorious for razor-thin profit margins.

A viable solution to both industries’ conundrums could be artificial intelligence.

Japan Data Science Consortium Co. (JDSC), a startup incubated at the University of Tokyo, believes it can solve this growing issue using its own AI patent that analyses household electricity data to calculate whether anyone will be home to receive a package during a given time period.

In other words, the AI comes up with a delivery route for truck drivers based on the electricity data. If the electricity meter is up and running, someone is probably at home to receive a delivery package, meaning drivers won’t arrive at an empty home and have to re-deliver the package later.

Based on the success of a pilot study in 2018 that reduced the rate of re-deliveries by 90 percent, JDSC, in cooperation with Sagawa Express Co. and the University of Tokyo, will conduct an AI demonstration as early as this summer to set up optimum delivery routes that take into account power usage data acquired from smart meters.

“No other country has done this kind of experiment before. The U.S. doesn’t have problems with re-deliveries because packages are usually left outside homes, and leaving packages is comparatively accepted in Europe,” JDSC’s Chief Data Science Officer Shimpei Ohsugi said in an interview. “Japan is one of the leading countries on smart meter installation and Japan has problems with re-deliveries, so that’s why Japan has become advanced in this area.”

The power data can be used for other services, including those that help relatives and caregivers remotely check on the well-being of elderly people living by themselves.

Smart meters are due to be installed at all households in Japan by the end of March 2025, with Tokyo Electric Power Company Holdings Inc. set to become the first major utility to finish the installations at all of the 20 million households it serves in and around Tokyo by the end of March 2021.

Using such data inevitably leads to privacy concerns, but Ohsugi dismisses those worries, saying that it is only the AI that looks at the power data. The AI will not tell the drivers whether specific households are likely to be empty, and instead it will just issue a recommended order of deliveries. Data usage is also based on residents’ approval for this purpose.

If it works, it could be a savior for the transportation industry. The number of corporate bankruptcies in Japan triggered by the labor shortage rose to a record high of nearly 200 last year, with the transportation industry accounting for the largest share as it struggles to catch up with surging demand for e-commerce.

The graph on the right shows an optimal delivery route based on AI, with blue dots showing successful deliveries, during an experiment at the University of Tokyo's Hongo campus. The graph on the left shows the route devised by a human driver. Red dots indicate unsuccessful deliveries, and red lines show the route taken for re-delivery. | COURTESY OF JAPAN DATA SCIENCE CONSORTIUM CO. LTD.
The graph on the left shows the route devised by a human driver. Blue dots indicate successful deliveries and red dots unsuccessful deliveries, while red lines show the route taken for re-delivery. The graph on the right shows an optimal delivery route based on AI during an experiment at the University of Tokyo’s Hongo campus. | COURTESY OF JAPAN DATA SCIENCE CONSORTIUM CO. LTD.

Re-deliveries have cost the parcel delivery industry some ¥200 billion per year in wasted fuel and labor costs — roughly 1 in 5 deliveries are delayed by absent residents. In fact, data show that a quarter of mileage by delivery drivers is spent on re-delivery and the issue causes a total manpower loss of 180 million hours a year.

Another parcel delivery giant, Japan Post Co., has been experimenting with AI algorithms developed by Nagoya-based startup Optimind Inc. to set up optimal delivery routes. Japan Post’s general manager, Yoshihiro Gomi, says the system frees drivers from the menial task of planning routes, allowing them to concentrate more on driving safely and allowing novice drivers to perform at the same level as more experienced drivers.

AI is being adopted in one way or another in almost every industry. The global battle for dominance in the field has been waged between the United States and China, with Japan falling far behind in terms of patents acquired. In order to catch up, the Abe administration last June set an ambitious goal of fostering 250,000 AI experts per year to strengthen industrial competitiveness and become a front-runner in AI applications to various industries.

In line with that goal, mobile phone carrier SoftBank Corp. teamed up with the University of Tokyo last December to set up the Beyond AI Institute, funding ¥20 billion over the next decade to accelerate advanced AI research in Japan.

“Japan is far behind the world in AI, but we will catch up and overtake them,” SoftBank CEO Masayoshi Son said.

Teruo Fujii, the University of Tokyo’s executive vice president, says the university has a global edge in basic research such as mathematics and robotics and added that he hopes the institute will help advance Japan’s applied AI research to the next level.

Though advanced AI is primarily being used at large companies, it is also finding a home at small to midsize businesses as well.

Ebiya, a long-established Japanese restaurant in Ise, Mie Prefecture, has developed an AI system that predicts the number of dining customers it will have with more than 95 percent accuracy, which has helped triple productivity, grow the company’s profits fivefold and reduce food waste.

The system, which predicts the expected number of customers up to 45 days in the future using more than 100 different sets of data, incorporating weather forecasts and the number of pedestrians likely to be by the restaurant, is so successful that Haruki Odajima, the president of the company, has made the system into its own business. The firm, EBILAB, markets the AI system to other restaurants for a monthly fee.

A cook at Ebiya prepares for the next day based on an AI projection that predicts the number of expected customers with more than 95 percent accuracy. | COURTESY OF EBIYA
A cook at Ebiya prepares for the next day based on an AI projection that predicts the number of expected customers with more than 95 percent accuracy. | COURTESY OF EBIYA

Data-driven management is still rare in the industry, where about half of new restaurants go bust within two years, said EBILAB’s chief technology officer, Ryuji Tokiwagi.

Odajima says that the idea for the system came following his restaurant’s failure to estimate the number of customers based on intuition and experience.

The benefit of accurate predictions is clear. Ebiya reduced food loss by 70 percent under the system, helping it optimize its preparation of dishes for the next day.

Currently, nearly 120 restaurants have introduced the system, with many turning around their businesses and succeeding in slashing food waste. More than 1,000 restaurants are considering using it and EBILAB is finding it difficult to keep up with robust demand.

“Restaurants account for the biggest share of business startups in the world, and yet most of them close down within 10 years,” Odajima says. “The restaurant business is often said to be the least productive industry that offers the lowest pay. I want the service industry across Japan to use this system and turn around this industry.”

Haruki Odajima | COURTESY OF EBIYA
Haruki Odajima | COURTESY OF EBIYA

After quitting his job at SoftBank, Odajima decided in 2012 to take over Ebiya, an established enterprise going back more than a century that was run by his father-in-law. He overhauled the restaurant, which previously used an abacus to tally up customers’ bills, turning it into one with a modern artistic design that is equipped with modern technologies such as sensors and cameras.

Last year, for a second year in a row, Microsoft named Ebiya as one of the most successful companies that utilizes AI. Odajima is now enjoying the spotlight as he gives about 200 speeches a year throughout the country on his innovation.

His next goal for his AI business is the global market.

“There are countless possibilities for the business, and that’s what I have learned firsthand,” he said. “Now I am thinking of doing business abroad as well.”

Coronavirus banner