FLYOVERS AID AGRICULTURE Kenji Shimada, professor of mechanical engineering, and his team of engineers are using drones to detect damage in agricultural water canals in a town in Niigata, an agricultural district on the northwest coast of Japan. These canals that total approximately 40,000 kilometers throughout Japan are essential for the rice farming economy. Damage to the canals accumulates due to age, earthquakes, and extreme weather. They can only be analyzed and repaired during the annual two-month dry season. Of these two months, one and a half are devoted to laborious inspections by technicians who walk along the canals to manually identify, measure, and record damage. This leaves only two weeks for repairs. “Forty thousand kilometers are equivalent to the equatorial circumference
of stone exposed, so these were the places
subgraphs, which represent the areas they
of Earth, and the manual labor for
we needed to repair.”
will cover. The team also graphs the roads
Using public maps and research data,
to generate a route for the car that is within
water canals is enormous. We automate
the researchers formulated an algorithm
communication distance of the drones. The
the work by flying autonomous drones
to plan the drones’ path along the canals.
cars are programmed to automatically find
equipped with high-resolution cameras
They can fly along different sized canals in
alternative routes if faced with traffic.
that detect cracks and wear with machine-
multiple directions to record video of the
The team presented their research at
learning algorithms,” said Shimada.
walls for crack detection. The commercial
the International Conference on Intelligent
drone they used is limited to thirty minutes
Robots and Systems (IROS) last October.
a systematic framework with a fleet of
in the air and it must stay within a range
In the future, they plan to tackle potential
drones and cars to effectively assess the
of a few kilometers from the remote
road blocks. Additional challenges to this
canals, extending the coverage area and
controllers. This makes it impossible for the
project include potential vehicle collisions,
minimizing inspection time. Di Deng, a Ph.D.
drones to cover all of the canal in one flight.
aerial constraints (such as flying zones,
candidate in mechanical engineering, works
To ameliorate these limitations, the drones
aerial traffic, and other regulations), and
on the coverage planning aspect of the
are paired with cars that are strategically
inaccurate maps and measurements. A
project. Last year, she traveled to Japan to
parked to provide batteries and pick up the
long-term goal is to develop large-scale
conduct field tests.
drones when needed.
automated vehicle planning.
inspecting and evaluating the condition of
Shimada and his team have developed
“Over the summer, we flew our
Once the drones have recorded video
“We believe that this type of technology is
autonomous drone and tried different
of the entire canal, the data from the
critical to keeping the aging infrastructures
sizes of water canals, so the system can
images is fed into a neural network to
healthy and safe—it enables faster,
automatically decide the position of the
detect damaged areas and map them in
cheaper, and more regular inspection and
drone inside the water canals,” said Deng.
CAD models. To plan paths for the drones,
monitoring,” said Shimada.
“We tried out canals that go from 2.4 to 5
a scaled map of a canal is converted to
meters in width. We could clearly see a lot
a graph. This graph is then divided into
