Fruit growers and labourers do not seem to be under immediate threat of a robot-led overhaul of the horticulture industry.
Hold tight - we’re checking permissions before loading more content
Tatura Smart Farm hosted a symposium of international scientists developing the frontline of research into fruit-picking robots on December 3 to 8.
The consensus of delegates at the event was that the release of the world’s first viable fruit-picking robot was not imminent.
Tatura Smart Farm researcher Ian Goodwin said there were several groups around the world researching robotic harvesting.
“I was pretty impressed with what I have seen,” Dr Goodwin said.
“But one thing with an agtech company is that they can start with a lot of drive in the actual technology but then it can go broke. It’s a cutthroat business.
“At the same time all machines are doing the same thing so it’s not a race, and the companies collectively share the technology as it comes along.
“But if you can get a large robotic business where agriculture is just a small component of their research, that’s where the advances will come.”
Dr Goodwin said the Penrith-based start-up company Ripe Robotics had established a research station at McNab Orchards in Ardmona to develop their robotic innovation.
“But that company depends on crowd funding.”
Ripe Robotics chief executive officer Hunter Jay said funding remained an issue for the company.
“I broadly agree about the financial challenge,” he said.
“We would love to have raised a lot more money and scaled it a lot larger, but we are doing what we can with what we’ve got.”
Mr Jay said the design to his company’s prototype — named ‘Eve’ — would still require changes from its closed-linked limb system to a motorised pivoting elbow.
“The system allows us to have all the weight at the back and the elbows are free moving pivots but it is not robust enough if it gets bumped and loses its calibration,” he said.
Dr Goodwin said the main issue facing robotic harvesting was the three-dimensional architecture of typical apple and pear trees making it difficult for robot arms to stretch around.
“The traditional branching of the pear and apple tree will need to go, with growers adopting the ‘wall’ structure for ease of access.”
Christian Andergassen from the Laimburg Research Centre in Italy was a key contributor to the symposium and agreed that tree structures needed to change drastically.
“For this to work, we need to adapt our orchards to the machine and not the other way around,” Mr Andergassen said.
“The key step is this change being made.”
The speed of the robotic arms shown to the delegates in videos allowed for one piece of fruit to be picked in around eight seconds.
Dr Goodwin said that rate needed to improve to about one piece every one second for a robot to be viable.
“Once we had a camera on the actual picking hand and this kept things slow, but the technology has now advanced so far that the camera can be mounted on the robot.
“That level of — I’ll call it artificial intelligence — is not the limitation.
“It’s the speed of the picking arms — it’s the actual mechanism of the arm picking the fruit that needs to improve with some speed.”
Mr Jay said his company was confident that it would be able to reach one piece of fruit per second on each four-armed robot.
“One apple every two seconds makes it commercially viable and there is no reason why you can’t get each arm to pick one per second.
“However, we are more focused on getting it reliable right now, so we are not looking at speed yet.
“We are more concerned about picking without fruit damage and having it not break down; we notice things like the arm wobbling a bit and it needs to focus on the fruit better.
“We last made a jump from 20 to 12 seconds and only took one or two weeks to make that work. There are a few simple things to speed it up.”