Page 12 - 11 Cotton SA March 2017

Robotics in Agriculture

Robots could scoop up the remainder of jobs in agriculture over the next two
decades, as machines become smarter, cheaper, and more efficient. That's
according to a new report by Lux Research that reveals how robots and
autonomous systems could take over and change the agriculture industry.

he term, robotic agriculture, may with large-scale market introduction Professor Perez said Agbot II has
Tsound very unreal and people may largely delayed not by technical issues demonstrated an outstanding
have different views about it. However, but by regulation, high sensor costs and performance in the use of robotic vision
it is a reality and soon it is going to be the lack of farmers' trust. This is and artificial intelligence for the
on every farmer’s lips. Robotic expected to change by 2022 when detection and classification of different
agriculture is a new way for farmers to sales of unmanned tractors are weed species. "The cutting edge robotic
use their implements without any labour expected to pick up. vision gives Agbot II the ability to spot-
and where the tractor operates on its spray selected weed species and use
own. Robotic agriculture is not restricted to mechanical tools to remove other weeds
tractors, but can be applied for any species that are herbicide resistant,"
Lux argues that in the near future, we farming need you have. For example, Professor Perez said. "To date, we have
are going to see farming robots that are a fully-autonomous agricultural robot concentrated on the three weeds that
not only much cheaper than human named Agbot II, designed and built by are relevant to Queensland: volunteer
labour, but capable of executing tasks the Queensland University of cotton, sow thistle and wild oats, and the
with much more efficiency and Technology (QUT) with support from vision system operated with 99%
accuracy. That could lead to higher crop the Queensland Government in accuracy in the classification of the
yields for farmers and a workforce that Australia, was demonstrated for the first correct species based on the images
doesn’t need to rest every few hours. time last year. It is estimated that this collected by the robot cameras." He said
Field crops are already seeing technology could save Australia's farm one of the key problems is that weeds
autonomous systems enter the sector $1.3 billion a year by reducing are becoming increasingly immune to
workforce. An Autosteer system for

Case IH Autonomous Concept Vehicle

tractors and harvesters has already the costs of weeding crops by around chemical control and that was why it was
reached a 10% market penetration in 90%. important an agricultural robot could not
the USA. only detect, but classify the weed
Professor Tristan Perez, leader of species on the spot and decide which
Tractor guidance and autosteer QUT's agricultural robotics program, actions to take to treat them.
technologies are going mainstream said Agbot II's sensors, software and
thanks to improvements and cost other electronics enable it to navigate "Agbott ll’s vision system can identify
reductions. More than 300 000 tractors through a field, detect and classify weeds and decide in real time which are
equipped with autosteer or tractor weeds and then kill them either better to spray and which are better
guidance technology have been sold mechanically or chemically. The robot dealt with, for example, mechanical or
globally in 2016 alone, rising to 660 000 can also be used to apply fertiliser. thermal methods," he said. "The light
per year in 2026. These tractors use weight of AgBot II, which is about 600kg,
RTK GPS technology to autonomously "In future versions, these robots could will help reduce soil compaction that
follow pre-planned paths with also provide feedback data on such affects the yield by limiting the root
centimeter-level accuracy. This makes things as soil and crop health and the development of the crops. Also due to
agriculture the largest adopter of state of diseases as they conduct their weight, the robots can be deployed
autonomous navigation. operations. This would enable better faster onto fields after rain to keep a tight
management decisions driven by real- control of weeds before they drop their
Unmanned autonomous tractors have time information," he said. seeds".
also been technologically demonstrated
12