Visual AI applications based
on teaching a camera system
to recognise objects, on the
other hand, are garnering
lively interest. “An AI-powered
camera system can take a single
image and make more decisions
in a split second than a human
operator,” Gesing enthuses.
“It can take measurements of
a shipment, determine if an
order is complete, if a shipment
is stackable, if it needs special
handling, and a host of other
things.”
Autonomous robotics
Meanwhile, in the warehouse,
autonomous robotics are
rapidly coming into play.
Gesing distinguishes between
moving robots and stationary
units. Currently, a lot are
used for vision-assisted
picking. For a robotic arm
that puts the stream of
shipments on a conveyor in
the right positions to feed
into a sorting system, every
move is a different motion,
which requires a self-learning
element. By the same token,
robots that move between
locations to pick items need
self-learning to have a sense of
their environment, explains
Gesing. “You can’t physically
programme all the possible
routes,” he acknowledges.
All of this requires
a sustained effort and
investment. “We don’t jump
into the pool and jump
back out. There has to be
continuous investment in
this,” comments Strauss.
Air Canada’s fi rst AI project
began with painstaking
work to prepare the ground,
building the infrastructure to
pool all the available data and
make it available to all the
tools it uses.
Augmented reality yet another strand in the web
from which the cargo sector can clearly benefi t
The fi nancial consideration
The fi nancial side remains one of the bigger hurdles for AI
adoption. APQC says that the major obstacles are securing the
fi nancial resources, fi nding people to perform the analytics
activities and obtaining the buy-in from top management.
Without all this, such efforts will not yield results, it warns.
Air Canada benefi ted from being picked as one of the initial
participants in ‘SCALE AI’, a supercluster supported by the
Canadian government that aims to build the next-generation
supply chain and boost industry performance by leveraging
artifi cial intelligence technology.
Gesing fi nds that it’s important that business and governments
work together on AI. “You have to have the right regulatory
support to make it work,” he explains.
This is a good time to garner state support for AI projects.
Gesing points out that more than 80 national governments have
come up with AI strategies. “They want to invest; they see it as
important for the future of their country,” he notes.
Air Canada is continuing with its exploration of AI. Besides
the project on spot rating, it is also working on pallet utilisation
this year. As far as Strauss is concerned, this is only the beginning.
“We see many more to come,” he says.
A paradigm shift?
Some onlookers predict a paradigm shift that will fundamentally
change the nature of the industry. They foresee a heavy emphasis
on predictive and prescriptive mechanisms that will turn the
game from a more reactive to a proactive model.
Strauss envisages a far-reaching change of his outfi t. “We’re
going to become a tech company that moves air cargo,” he
declares succinctly.
“Usually, in the initial
period, you spend a lot of time
co-ordinating and classifying
data so it is available for use,”
remarks Strauss. The carrier has
about 92m data points, which
the system checks multiple
times a day for updates.
“IoT is a huge focus for us,”
remarks Gesing. “It generates
a lot of data that is hard to
use without AI. It’s also very
diverse data.”
Air Canada has set up a
laboratory where possible
solutions undergo checks for
proof of value before they
are tested in the real world
in a limited environment. It
has brought in data scientists
from McKinsey and has
created a permanent role in
the organisation to look after
this area.
The cost, though, is
considerable. “This is a multimillion
dollar tab. It’s fairly
complex and expensive, but
we think it creates value,” says
Strauss of the investment.
There has to
be continuous
investment
Tim Strauss, Vice President Cargo,
Air Canada Cargo
www.airlogisticsinternational.com April 2020 35
/www.airlogisticsinternational.com