25 May 2018
What John Deere, the vast US maker of farming and
construction equipment, means by "Where’s my
tractor" is: "Where is it to within 2.5cm?"
This was one of the gems in a conversation I had at Mobile
World Congress (MWC) in Barcelona in February with John Stone,
an SVP at the company.
But there was more in this conversation about where the
internet of things (IoT) is really going. It’s all
about precision agriculture, and that means not only getting
your self-driving machine to cover every inch of the field,
right up to every wiggly boundary, but to place seeds just
where they’re needed, just at the right depth with
the right soil contact, and not twice in the same
Wiggly boundaries – my term, not
Stone’s – are not a problem in the wide
open plains of the Mid-West (look out of your airliner window
next time you’re flying over) but they are in
Europe. Take a walk just about now to the bluebell woods of
Kent, in south-east England, and you’ll see what I
mean. You can miss a lot of field at the edges, even when you
are an experienced tractor driver.
Spot the weeds
And what else does Deere’s equipment do? It
makes sure you never spray the crops twice –
especially if spraying isn’t needed. The
traditional method has been to cover everything with
weedkiller, just to make sure you zap all the weeds.
That’s wasteful, said Stone. "We can see the
plants and we can see the weeds."
The technology comes from Blue River Technology, an
artificial intelligence (AI)company based in Sunnyvale, at the
heart of Silicon Valley, that John Deere bought last year for
$300 million. Jorge Heraud, co-founder and CEO of Blue River,
explained the logic behind the deal last September: "Blue River
is advancing precision agriculture by moving farm management
decisions from the field level to the plant level," he said.
"We are using computer vision, robotics and machine learning to
help smart machines detect, identify and make management
decisions about every single plant in the field."
And that’s what’s happening today,
Stone told me in Barcelona this year. "In the last few months,
with AI, we’ve trained the system to identify
weeds and plants more precisely than a trained agronomist can.
It means we don’t get herbicides on plants that
It means the farmer is managing each field down to the level
of the individual plant. Or weed.
And it’s not just in planting the seeds and
dealing with the weeds. Several months later, when you are
harvesting the crops, you know the truck that follows the
harvester, just to one side and just behind, catching the grain
that’s sprayed out? That requires careful driving
to get all of the grain. A lot falls, wasted, back to the
The IoT alternative, thanks to John Deere’s
technology, is a self-driving harvester and a self-driving
truck: they stay together automatically, just in the right
place. "With GPS guidance you get a payback in one season,"
said Stone. The technology that allows the farmer to seed every
wiggly edge of every field also delivers a payback in one
season, he adds.
Big farm data
The whole system is really an agricultural big data system,
an edge computer trundling through the fields, with a choice of
attached planting attachments, weeding attachments and cutting
attachments, complete with Bluetooth and Wifi, gathering data
on what seeds were planted and what’s
But out in the countryside, I hear you screaming, how do you
beam this all up for processing? There’s precious
little broadband coverage in the wide open spaces of Utah or
Ukraine. Or Kent, if you’re honest.
"We have telecoms partners that can install towers for large
farms," Stone told me. John Deere is working with AT&T for
the telecoms, AWS for the cloud, as well as Nvidia and Intel
for the edge computing, he added. "If there’s no
coverage, you store and forward."
But this is clearly an opportunity for carriers: 4G
– even private LTE – in and around the farm.
And lots of opportunity for fibre providers to deliver backhaul
networks to those 4G towers and fibre direct to the farm
Bandwidth and speed
John Deere has already delivered over 100,000 large machines
with 3G and 4G for wireless data, and Stone is looking forward
to 5G. He told me: "The bandwidth and speed promises of 5G will
be very important. It means we can do more."
Look what the regulators and wireless carriers are talking
about: making spectrum available down in the 600-700MHz range.
As any radio engineer will tell you, with a nod to work done by
Claude Shannon at Bell Labs in the 1940s, the available
bandwidth won’t let you carry very much
But the signals carry a long way – think of those
UHF television stations that were formerly on that bit of
spectrum, delivering colour pictures across whole counties.
Perfect for agricultural IoT for Utah, Ukraine and even
Sounds expensive. "Yes, but farmers buy things only where
they can see benefits quickly," he said. Already
Deere’s internet of weeds cuts herbicide use to
10% – that’s to 10%, not by 10%
– of what it was. And the farmer saves on seeds and
fuel as well as chemicals. "It’s precision
agriculture. You save costs, and you grow your yield," he