Engineers from Monash University have developed a robot capable of autonomous apple picking that could help fill the labour gap caused by international border closures.
The robot can identify, pick and deposit an apple in as little as seven seconds and, down the line, could provide a solution to labour shortages in the agricultural sector.
With borders still closed due to COVID-19, Australia’s fruit and vegetable farmers are facing a shortfall of 26,000 workers and incentives from federal and state governments are yet to entice enough workers to fill the gap.
Dr Chao Chen, from the Department of Mechanical and Aerospace Engineering at Monash University, who led the team behind the robot, said he was excited to see it come to life after about three years’ work.
“There has been an increased demand for automation in Australia,” he told create. “Now, with the shortage of backpackers at the moment, there is an opportunity to use technology to fulfill that need.”
Seeing and touching
For many years, human apple pickers have had the advantage over robots due to the simple fact that they can see what they’re picking. To develop an apple picking robot, it had to be able to identify apples from the leaves and branches around it.
“The robot has a RGB-D camera which takes photos of the tree. It then uses deep learning to detect the apples and their location on the tree,” Chen said.
“After the detection, the robot finds the best path to approach the apple to pick it.”
According to Chen, the robot can correctly identify more than 90 per cent of the apples within its view in less than 200 milliseconds. It can also work in different lighting conditions.
“It actually works better at night,” Chen said, “so even though it’s slightly slower than a human, it makes up for the time difference by being more versatile.”
A human can pick an apple every four to five seconds, while the robot takes roughly seven to ten seconds.
Once the robot locates its target, it moves its arm towards it and picks it with a soft gripper equipped with a suction cup.
“The system receives the information and identifies the optimal sequence to harvest the apple. The path is planned accordingly to harvest the apples from the tree and drop them into a collection box,” Chen said.
The choice of soft fingers for the gripper means the robot won’t damage the tree or the apple during the process. The suction system within the ‘palm’ of the gripper allows the robot to pull the apple away from the tree, further minimising the potential damage to it.
Part of the robot’s success has been its ability to keep the apple’s stem intact when picking the fruit, which is a cosmetic requirement of some retailers.
From fruit to vegetables
To truly assist with Australia’s labour shortage, Chen hopes to apply the robot’s skills to other fruits and possibly vegetables.
“We will need to customise the technology so it can identify things like pears or oranges, that’s probably the next step,” he said.
“Currently our robot moves and harvests from a vertical plane. To harvest vegetables from the ground is a different orientation, but we are confident our technology could be applied there.”
The team also plans to add more versatility to the robot by expanding its ability to scan a tree for any growth problems or wayward branches.
While the robot has the potential to increase the productivity of farms, Chen also hopes the use of new technology could draw younger people to the agricultural sector. As Australia’s population grows, food production is only going to become more important and there is a potential for engineers to play a greater role in the process.
Took more like 15 secs on average!
Congratulations.
Could you build the grab arm around a tube that could then roll or suck the apple into the hopper. Thereby reducing the travel and unravelling time. Grab apple, roll, pull off, let go (apple rolls down chute), grab the next. Chute could be designed to minimise bruising. Then with the reduced movement you could have two arms working together up and down the trellis.
Just a thought without any experience
Cheers Noel Hobley.
Would be quicker by hand!
Were I an Orchard Ganger and heard fruit dropping into a bin/container ( as are those apples) I’d be very quickly telling the offending Picker to take more care of the product!
A start is a start…I am certain it get more sophisticated as time goes on…and surely it cannot be persuaded into ridicules jokes of some of the orchard managers…?????
We do not need this technology in fruits picking, to keep the price of fruits low and for efficency. The problem of shortage of labor could be solved well by more practical government policies and good wages to attact people especially who have no jobs to work in picking of fruits.
Yep but the robot would work non stop unlike a picker
I guess it’s a start, but the fruit being dropped bruises them, and they will all need to be thrown out. Apples need to be carefully placed on each other when being picked.
Could a farmer afford a robot like this ?
This should’ve been used in the garden of Eden. The apple would’ve been picked and not eaten.
Too much damage to the tree when twisting and pulling. Would be better if it had a cutting tool at the tip just after the apple. No twisting. And besides how do you have a conversation with a machine? No , better if we stick to humans , hands are more sensitive. When we work together in groups we interact and the fruit tastes better and sweeter!
A rough start to fruit picking, apples should be placed gently into container and not dropped.
These trees need to be hand pruned and hand trellis trained each year for this machine.
I have previously picked fruit and vegetables and pruned trees and grapevines for 20 years. This gave me the inspiration to study and become a certified horticulturalist.
The best way to attract fruit and vegetable pickers and pruners is to pay good money. Less unemployment! The last time I pruned grape vines in 2010 I earned $25 per hour. The next year, 2011, foreign teams accepted the offer of $11 per hour to prune grape vines.
Big super markets should pay the proper $ for produce.
It is very encouraging to develop an Australian prototype robot for apple picking. The concept of using robots in fruit harvesting is not new. Research indicated that robot technology is not commercially sound because of some challenges:
1- fruit damage
2- fruit handling
3- fruit losses (clearance rate)
4- picking speed
5-tree damage (broken branches)
6- technology cost
7- harvest cost
8-robot maintenance cost
9- ease of use.
Therefor there is a need for a research team funded by the Australian Fruit Industry to tackle the above challenges and come up with a complete robot harvesting system which includes picking, transferring, packing and handling. As far as I know, the Australian fruit industry used to be in favour of robotic technology and they preferred harvest aids.