Australian researchers branch out with robot farming

More News Most Read Today's Headline
Australian researchers branch out with robot farming

Some orchardists were skeptical when the idea of robot field trials was proposed by Dr Salah Sukkarieh and his team at the University of Sydney, who sought to detect key agronomic inidicators like yield, tree health and flowering intensity.

A wide range of benefits to the technology have been shown in the first two years of the project funded by Horticulture Australia Limited (HAL), but Sukkarieh's ultimate goal is to solve Australia's pressing issue of labor shortages through automated harvesting.

"After the first year when we'd demonstrated all these things they [orchardists] were quite over the moon. Automatically what they see is not just a crop yield, not just the detection of the fruit but the potential behind all of that," he told

"At the apple farm we could go down a row, count the number of apples and could tell the farmer the crop yield. That was something they used to do once before but it was too laborious.

"The fact that you could just take the robot up and down the orchard for an hour and get a crop yield was something phenomenal for them."

Photo: Australian Centre for Field Robotics at the University of Sydney

Photo: Australian Centre for Field Robotics at the University of Sydney

In addition to the apple field trials in Melbourne, the team has also helped growers at an almond operation in Mildura.

"Last year we had detected the almonds on the tree with the robots, but this year what we decided to focus on was the flowers," he said.

"Just before the fruit set, the almond trees blossom and what farmers would like to know is the intensity of the flowering, because that governs a lot about what they expect in terms of fruit set - that tells them how much water and fertilizer they need over the next few months."

With one year of the current HAL funding remaining, Sukkarieh hoped to explore the technology's potential on mango and banana farms in the country's north in late 2013.

He added that from a field intelligence perspective, robotics experts had already done a great deal of work in monitoring invasive species.

"There's been a lot of work in using robotic aircraft and intelligence surveillance systems and algorithms to detect invasive weeds and locus tracking, and they're going to be applied to this horticulture project.

"We can detect weeds on the ground, we’ve got high resolution sensors so we can detect insects on the trees, but there are other aspects such as the health of a tree so that’s a little bit harder. You need a bit more work and interaction with plant biologists."

A path to automated harvesting

Sukkarieh said the robots, named Mantis and Shrimp, used a range of sensors to conduct field intelligence including laser vision, radar, thermal and infrared, while the team had also borrowed conductivity and gamma sensors from the university's precision agriculture group to measure water properties and ions in the ground.

Salah Sukkarieh

Salah Sukkarieh

As these two robots were originally built for general purpose use - particularly in areas like mining, defense and urban mapping - Sukkarieh expected the end product would look quite different for farmers.

"You could actually use those sensors now and put them onto a tractor for example, and while you're driving that you can just be collecting the data and looking at various metrics.

"Obviously the desire is to go down the automation path because you can get much higher precision and 24/7 operation, but they can already use the information that we’re deriving.

"To be able to harvest you first need to be able to detect the fruit."

The expert said robot harvesting could be a reality within "a couple of years" if funding were no issue, but it would most likely only be suitable for certain types of orchards.

"If you look at this tree architecture that’s popular in an orchard, very much what you see is a tree and a bush canopy; these are 3D kind of arrangements and those are very hard because no matter what the fruit is, you can only harvest what’s on the outside.

"To reach in to harvest what’s on the inside is going to be impossible, robotically anyway, for a long time."

He highlighted there had been a great deal of activity in changing tree shapes for biological purposes, and these same set-ups  were also favorable for robot harvesting.

"It’s been found out that if you grow certain crops on 2D trellises and you have these trellises turned at a certain angle to light then you get optimal photosynthesis and the most efficient type of tree, and you get to improve crop yield.

"In those types of situations the fruit is hanging on one side of the tree only, and that will open up a lot of opportunities for automation.

"These structures are already being used with apples for example, so you can imagine the apple industry will probably be one of the first to see automated harvesting happen."

From mining to "dining"

In July, Australia's Federal Agriculture Minister Joel Fitzgibbon coined the term "dining boom", following on from the mining boom that has kept the nation relatively prosperous through the global economic crises of recent years. Government strategists, industry and researchers are set on making the country Asia's "food bowl", and Sukkarieh's work is emblematic of this priority.

Having undertaken robotics projects with mining multinationals like Rio Tinto and BHP Billiton, Sukkarieh said automation in horticulture would be very different due to the unique nature of farm operations.

"It's all about money. Mining can afford to spend a lot of money on R&D and can afford to put expensive sensors on trucks and actually see automation work properly.

"You're talking about large-scale operations that are run by a company, whereas in agriculture - especially in the orchard industry - you’re talking about smaller holdings, smaller farms, machinery that can't be expensive and R&D funding that is nowhere near as great as what you’d find in mining or defense.

"The challenge is being able to most effectively deliver some sort of automated system given the low level of R&D with something that can be used on the farm practically, robustly and inexpensively."

As has been the case in Australia's mining industry, Sukkarieh emphasizes there simply aren't enough workers available.

"There are two ways of getting around that - one is to increase your immigration and you dedicate certain visas to do that, or you look at automation to help you along the way.

"In Australia there's been a focus on automation. We have a proud history over the last 20 years – we have some of the largest automation projects in the world happening, and agriculture’s just going to be another one."




Subscribe to our newsletter