Articles, Blog

Technology meets agriculture | Hiroshi Mineno | TEDxShimizu


Translator: Masako Kigami
Reviewer: Claire Ghyselen We all know that
the Japanese society is aging. But where exactly
does it stand in its transition ? This is a graph from a white book
on aging society issued by Cabinet Office. The Japanese population
has peaked in 2015. As statistics show that
the total population would be decreased due to the declining birth rate
year by year. In 40 years, in 2055, the number
of 65 years old or more in Japan will be increasing to reach 40%
of the total population. In other words, in 40 years,
our generation will be the aged one. Now, do you know
how many people are active in the first industry,
agriculture, forestry and fisheries? The number will be decreasing
in spite of aging year by year. Currently, the first industry account for 3.6% of the Japanese economy. Will it disappear from Japan
in the future? Let’s have a look at
agricultural production The upper graph shows
the gross agricultural output. It is decreasing every year, reflecting to the decline of
the number of farmers. Therefore, the income
of production agriculture is decreasing year by year, too. The image of agriculture
is strongly connected with hard work and unstable income. If they are
stability-oriented young people, they won’t venture into farming. There are three consequences to this: ageing is getting faster, the number
of farmers is decreasing fast and untended arable land is booming. I think Japan is stuck
in a viscous circle. Do you think future of the first industry
in Japan should be miserable? I don’t think so. Most of our generation
was not educated into farming. But we are used to diligently working
from morning to night for companies. Some of you may long
for reconnecting with nature, because it allows you
to work at your own pace. There are many possibility
to increase production and profits by improving production processes based on information
and communication technologies. Here is my strategy. Produce local
special products in the first sector. Next, add add high value or branding
in second industry. And then, distribute and sell effectively
in third industry. 1 x 2 x 3=6 This is the sixth industry: a complete industrialization
of agriculture. Of course, there are issues related
to food safety and risk management. However, honestly speaking,
we should be over pessimistic when facing with a super-aging society. By the way, do you know
the technology of “plant factory”? I’m sure some of you know about it. When searching “plant factory” on Google, the results are as follows: There are two kinds of factories. The first one is based
on natural light from the sun. The second is based on artificial light
such as LED and fluorescent bulbs like LED and fluorescent bulbs. Today, major electronic makers
renovate a cleanroom into plant factories
with LED or fluorescent light. They produce lettuces and leaf vegetable
and sell them. As I’ve just mentioned it, we are moving towards the sixth industry, the complete industrialization
of agriculture, How about other countries in the world? This is the concept of the “vertical farm”
developed by Dickson Despommier, emeritus professor of Columbia University. The meaning is “producing food
in vertically stacked layers” in Japanese. This concept is one of the propositions to use land effectively in urban regions. In a building like this one,
you feed farm animals, but not only. You also cultivate plants
such as leaf vegetables and fruits. You can then sell your products in the supermarket
located on the first floor. This system makes it possible to create
a circular agriculture within the building. If we can create
such a system in urban area, we can reduce
transportation cost drastically as well as create a bright future thanks to the creation of new jobs
and a more balance way of life. There are various challenging issues, like making the system economically viable
and getting a better energy efficiency. But when I saw this drawing by designers
based on the concept, It made me dream that someday,
I could make it possible. So I’ve studied day and night. Let me share my dream with you. Do you know the mechanism of plant growth? It is written on the science textbook
for fourth grade children. Photosynthesis takes place
on the surface of leaves and evaporation from their backside. That’s how plants grow. Photosynthesis is the process
used by plants to convert CO2 in air and water absorbed from their roots into organic matter, using sunlight. Evaporation from the backside
of the leaves starts when the sun rises. Water and nutrients
are absorbed by the roots and are conveyed to their whole organism. If we can measure their growth, it should be possible to control
their rate of growth, shouldn’t it? How do I approach this simple question? The growth of plants is connected
to photosynthesis and evaporation, and both processes are located on leaves. So, if we can count indirectly
how many leaves a plant has, we should be able to measure its growth. Now, I am sharing the machine I developed. I’ve named it a “wireless
scattering light censor”. To grow, a plant needs
photosynthesis and evaporation. This process is related
with how many leaves a plant has got. The more leaves a plant has got,
the darker its stem is because leaves grow to catch sunlight. That was why I developed this wireless
machine to measure light amount between its surface and stem. We can also measure temperature,
humidity and brightness. I also ponder how to estimate evaporation and photosynthesis
based on lush leaves. Water and nutrients that plants absorb
are important, too. As for hydroponic culture, by measuring the amount of water in a tank, we can deduct the volume
absorbed by the plant. If we do the same
for the nutrient, and measure them, we can deduct how much
was absorbed with the water. We developed various sencors and established an environment
to enable us to measure many temperatures, humidity, brightness
and concentration of nutrient solution. By combining those data
to statistical data prepared in our lab, we could logically deduct the actual amount
of nutrient absorbed by the plants. We then encoded this huge amount
of data into our computer. Based on current temperature, humidity,
water level, concentration and past data, the computer calculated
reasonable nutrient absorption amount. That is my challenge. We aim to reproduce the know-how
of professional farmers who are depending
on their experience and intuition. This is the core project. But I have a concern about the volume
of data we enter in a computer. Should we consider it as empirical value? The more is it really the better? We need time and efforts
to collect huge amounts of data. We used three-year worth of open data from the Automated
Meteorological Data Acquisition System to be learned by our computer. We compared the errors between the actual data
and our computer’s simulation. Japan stretches from North to South.
It is a very long country. So I picked up weather data for Sapporo,
Tokyo, Hamamatsu and Naha. There are 3 years of data. We’ve analyzed how errors evolve. Interestingly, for Naha the bigger data is,
the smaller errors are. As for Sapporo, Tokyo and Hamamatsu, especially Sapporo, big data is not connected
with better results. We obtain more exact results
when we use data for 8 months. According to this, we made a hypothesis there is appropriate data volume
to enter in a computer. However, it is impossible to encode
each regional data. and verify the best data volume
for each region. So I came up with a better way: The computer has extracted
various data randomly from big data. Based on this data, we build various
prediction models that our model is using. Let’s imagine what will be the temperature
tomorrow morning at 7:30. Each people will have
a different imagination. For example, it is spring now. “Morning temperature in spring
should be like this.” You estimate the temperature
based on the word of spring. But you can estimate it based on recent
morning temperature, too. “Tomorrow’s temperature will be like this
from the recent weather trend.” If you are good at statistics, you will extract similar data
from past data and analyzes them. You may think your estimate will be
more accurate using statistics. Anyway, we shall know
the accurate data tomorrow. After we know the correct answer, we will know which model
was the most accurate. So, we will continue
to make predictions with this model. If we notice that our current model
doesn’t make accurate prediction, we will use another model
and see how accurate the prediction is. This takes place automatically. I coined this algorithm as “Sliding Window-based Support
Vector Regression (SW-SVR)”. As you are here,
please remember the name, “Sliding Window-based Support
Vector Regression”. (Laughter) We work in an agriculture labolatory to conduct experimentations
in Iwata, Shizuoka Pref. We will collect environmental information, using this wireless
scattering light censor.” And we calculate
plants’ nitrogen absorption amount using the SW-SVR
during six hours in a computer. By controlling nitrogen supply
based on these calculation results, we measure out
just the right amount of solute. It is important to get
just the right amount because in case of growing tomatoes,
for instance, they will get bigger
by using such a solute. However, if the growing speed of its fruit
is faster than its skin, the skin will crack. So we can’t sell tomatoes
with broken skins. So too much or too less solute
is not good. It should be appropriate. This graph is one of the results
of last winter’s experiments. A black line shows
nitrogen absorption by plants. The red line shows
our algorithm’s prediction of the nutrient absorbtion. The black and red lines were almost same on November 26
with our first prediction model. However, as time passes, the difference
between them gets bigger. So we decided to re-create the model
automatically on December 20. Since then, the black and red lines
moved similarly. We can control them. Looking at these results, we see that computers can mimic
human’s know-how to some extend. At the beginning, I told you
not to pessimistic . about the super aging society We will establish networks
between these fields in difficulties, and mimic the know-how. I think we can create the 6th industry thanks to information and communication
technology. If we operate non-professional concepts
professionally, I think we can create a bright future
together. I would love to create
such a future with you. Thank you very much. (Applause)

One Comment

Leave a Comment

Your email address will not be published. Required fields are marked *