| Latest Articles | Category: Region & Industry |
Back number Search (FY2011- IIST e-Magazine):| Category Search | Keyword Search | |
Back number Search (FY2001-FY2010 IIST WORLD FORUM Mail Magazine): | Keyword Search | Timeline-Search |



IIST e-Magazine

Series: Employing AI/IoT to Create New Businesses (3) Bringing Road Heating to More Sites AI reduces running costs and concern Takuya Irisawa CEO Ecomott, Inc. [Date of Issue: 31/October/2017 No.0273-1062]

Date of Issue: 31/October/2017

Series: Employing AI/IoT to Create New Businesses (3)
Bringing Road Heating to More Sites
AI reduces running costs and concern

Takuya Irisawa
CEO
Ecomott, Inc.


Road heating is a great way of reducing the snow-clearing burden in areas with heavy snowfall, but given rising fuel prices, etc., the running costs of snow melting apparatus are frequently raised as a concern. Now a Hokkaido company has found a way to use AI to slash fuel costs. How did it manage to replace the “human eye” with AI?


Our Yurimott technology is a 24-hour remote monitoring service for road heating. An on-site surveillance camera that checks snow conditions is used alongside weather forecasts and other information, and the boiler switch is switched on or off remotely based on the results. This greater operational efficiency brings down the cost of the fuel used in road heating boilers—usually kerosene or gas—by more than 30 percent.

What is road heating?

Sapporo City is the only city in the world with a population of over one million people and more than 500 centimeters of snowfall. In other words, urban functions coexist with the major inconvenience of heavy snow. Snowfall in itself isn’t a problem as long as there is somewhere to put it; it can just be left there until the weather warms up again and the snow naturally melts away. However, because Sapporo is a city, there aren’t many empty spaces in residential areas where snow can be stored.

This has resulted in the installation of a lot of road heating, whereby hot water pipes are laid under the asphalt to create an outside version of underfloor heating that actually melts the snow. Aside from Sapporo, there is also a substantial amount of road heating in Otaru City and also Aomori City (although this isn’t in Hokkaido).

Entrance to a parking lot where road heating has been laid. The only areas clear of snow are those with pipes underneath.

 Entrance to a parking lot where road heating has been laid. The only areas clear of snow are those with pipes underneath.

Issues to date with “automatic” operation

Road heating has traditionally been automatically operated using snow sensors, which means that sensors play a very important role. When snow falls, the sensor reacts and automatically flicks on the boiler. However, sensor reactions aren’t very sophisticated, turning boilers on in response to rain as well as snow, or to brief snowfalls that end almost immediately.

Snow sensors also usually have a delayed timer that can be set so that even once the sensor detects that the snow has stopped, the switch will stay on for an additional 10 to 180 minutes according to the setting. This is designed to prevent large amounts of snow remaining on the ground even after snow has ceased falling. As a result, because the setting can only be changed on-site, if it is provisionally set at 30 minutes, the sensor may react to even very light snow or snow that melts away immediately and then leave the boiler in operation for at least 30 minutes.

Reducing fuel consumption with the human eye

Where sensor-operated road heating has therefore been quite wasteful, the Yurimott service sets out to resolve this problem by determining with a human eye whether snow is still on the ground and the boiler needs to be on. If the snowfall is so light that it melts away or doesn’t really pose a driving problem, a human being remotely monitoring the snowfall may decide not to turn the boiler on at all. This leads to big savings on fuel costs.

Cameras and IoT devices are installed on-site so that the remotely-operated boiler switch and the reactions of the snow sensor can be checked on a dedicated Cloud application screen. In Sapporo, snowfall can be completely different even in different parts of the city, so the operator looks at the characteristics of each particular area and combines these with weather forecasts as they work on the Cloud screen. Currently, our system monitors 1,800 sites primarily in Hokkaido and Aomori.

A camera installed on-site distinguishes the exact state of the snow in place of a human eye.

 A camera installed on-site distinguishes the exact state of the snow in place of a human eye.

Can AI do the work of a person?

Because Yurimott is a 24-hour remote monitoring service, having staff constantly on hand, including weekends and holidays, has until now required around 20 people. In addition, because Yurimott is a winter-only service, we have had to employ people every year just for that period. However, the recent social situation has caused major management difficulties, including personnel issues and rising wages. Having different operators on board every year has also meant shouldering the education cost of training the new staff each time, and sometimes operators who are just starting out turn switches on too early or too late.

These issues prompted us to look at whether we could use AI to do the work that traditionally performed by human staff. Our development efforts resulted in an AI-based system that anyone can operate. Basically, the parts of the process gauged by the human eye are (1) deciding whether or not there is snow, and (2) what the weather is likely to do next. If a computer could handle the necessary tasks of (1) recognition, (2) judgment, and then (3) decision-making, it would be possible to replace the human element with AI.

Multiple staff managing the 24-hour monitoring system

 Multiple staff managing the 24-hour monitoring system

AI strengths and areas where learning is needed

Starting with the screen image, determining whether or not there is white snow on black asphalt is far easier than human face recognition. The problem is black ice, when the asphalt looks black but is actually frozen. In this situation, obviously the boiler can’t be turned off. Our solution was to introduce an algorithm whereby the computer uses light reflection to determine whether or not the asphalt is frozen.

For the forecast aspect, an application programming interface (API) is used to acquire the current temperature and wind speed within a one-kilometer mesh, as well as local weather forecast data, from a weather forecast data transmission company. This data is then linked with AI via the Cloud. Image recognition and API linkage therefore enable AI to understand whether the switch should be turned on or off for the particular property.

Having AI handle remote monitoring in place of physical staff obviously makes things much easier, but there are some things that AI needs to learn. Looking at actual cases when AI decided that the boiler should be on, customers have sometimes come back and asked us to use the boiler only when snow would make driving a real problem; anything less, and they would be happy to clear away the snow themselves.

While a person will register these individual requirements and operate boilers accordingly, AI doesn’t cope well with such qualitative information. It needs to be taught that for this particular property, these are the conditions under which the boiler should be switched on. Because of this, the system is still not entirely run by AI; rather, the AI sends alerts to which human operators respond.

This IoT device used in remote boiler operation receives signals from snow sensors.

 This IoT device used in remote boiler operation receives signals from snow sensors.

The road to completely AI-operated monitoring

If road heating can become completely AI-operated by 2020, it will bring down remote monitoring costs, while IoT devices are also likely to become cheaper, which should open the way for remote boiler operation at more sites and help reduce carbon emissions.

As a company that creates social infrastructure by combining AI with IoT, Ecomott will continue to use the AI x IoT combination to solve social challenges, including not only carbon emissions but also disaster prevention, accident reduction, crime prevention and nursing care, developing services that will go on to become social infrastructure.

Application Programming Interface: Open-source software available to anyone for the development of new services and secondary data utilization, etc.


About the Author
Takuya Irisawa, CEO, Ecomott, Inc.

Takuya Irisawa, CEO,Ecomott, Inc.
Born in Sapporo, Hokkaido in 1980. Graduated from Highline Community College in Washington, America in 2002. Took an MBA at Otaru University of Commerce with a major in entrepreneurship in 2010. After returning from America in 2002, he entered Crypton Future Media, an IT company in Sapporo City, where he handled mobile phone content planning and development, including developing popular sites such as “Pocket Sound,” which has more than 10,000 paid members. He left Crypton in January 2007 and set up Ecomott, Inc. a month later. In June 2017, Ecomott was listed on the Sapporo Securities Exchange “Ambitious.” He is also currently Executive Director of the Hokkaido Information and Communication Technology Association, a member of the Human Resource Development Committee, Deputy Secretary of the Hokkaido Mobile Content and Business Council, and Deputy Director of the NPO Sapporo Biz Cafe.





| Latest Articles | Category: Region & Industry |
Back number Search (FY2011- IIST e-Magazine):| Category Search | Keyword Search | |
Back number Search (FY2001-FY2010 IIST WORLD FORUM Mail Magazine): | Keyword Search | Timeline-Search |