Beating labour shortages with robots might be a key to meeting the last government’s target of doubling export receipts by 2025, Waikato University researcher Professor Mike Duke says.
He oversees a department of bright young students working hard to take robotics outdoors, to the sharp end of the primary process where picking, planting, harvesting and thinning are all vital to crop productivity and earnings.
However, they were also the area’s most reliant on labour and the most susceptible to labour shortages and costs at critical times.
And seasonal challenges already being faced were only going to get worse.
Last year a Ministry for Primary Industries report estimated the booming horticultural sector needed nearly 8000 more skilled workers by 2025 to cope with predicted production rises.
Only last month Zespri said it was increasing its SunGold licensed growing area by almost 50%, adding 700ha a year for each of the next five seasons.
That doubled the fruit’s volume to about 100 million trays by 2025.
It also validated efforts already under way to automate picking, pruning, spraying and pollinating via robots that would begin next season.
“The challenge now is how can we move robotics into the paddocks and orchards?”
Duke was surprised by the often antiquated equipment some primary sectors used for specialised tasks and how ripe they were for a robotic revision.
He cited a pine seedling company in Tokoroa that tasked his department with making a robotic machine capable of drilling more precise holes for seedling planting rather than the off-centre holes a simple round spiked wheel created.
That machine’s round wheel shape meant a portion of the seedlings inevitably grew crooked.
Having done that his graduates also developed a machine capable of uplifting seedlings, removing dirt, grading them and boxing them up.
“Despite 200 people being unemployed in Tokoroa, this company simply could not get enough people to do this critical job.”
Software in the machine included algorithms replicating human decisions on each seedling’s root structure and proved to be as accurate as human graders.
The prototype was now operating and two grading machines would be running in the South Island by Christmas.
Tender Tips Asparagus was also highly reliant on seasonal labour for picking and wanted to replace more people with machines.
“It is back-breaking work with workers covering 600km for every 10ha. They are looking at robots capable of operating 24/7, picking asparagus as they appear.”
The biggest robotics project in the country was the automated kiwifruit robot developed by Robotics Plus with $10 million from the Ministry of Business, Innovation and Employment, Zespri and Robotics Plus.
With a robot platform already developed (Farmers Weekly March 27) the proof of concept had been demonstrated for picking and pollination was being tested this month in Bay of Plenty.
Duke was clearly excited by some cutting-edge technologies including LiDAR (light imaging detection and ranging) in the machines.
Also known as 3D scanning, the technology not only helped guide the vehicles but could also identify the ripeness and quality of individual fruit and make a picking decision.
Researchers could also see the robots becoming full time biosecurity sentinels as they picked, pruned or sprayed crops.
“We have placed replica stink bugs in the orchards for the robots to detect and they are proving very efficient at doing so.”
But the deeper, more challenging technology underscoring the machines was the same technology raising some questions among their designers and the public.
Many used deep neural networking or deep learning technology to learn their required actions without the usual prerequisite of coded input from a programmer to perform each specific task.
They understood and remembered patterns, learnt them and adjusted to variations. They were used widely for tasks like voice recognition, language translation and, increasingly, in self-drive vehicle technology and guidance.
But the technology was also regarded as something of a black box even by the engineers who built it, on grounds the complex networking or wiring of the systems’ artificial neurons and how they communicated were not fully understood.
As soon as the black box of AI was opened, the opportunities became endless but there was tension around the technology, Duke said.
“It is not such an issue with something like kiwifruit harvesting but if you took a self-drive car, for example, and it avoids another car and kills a pedestrian in the process then how do we understand how the car made that decision or where responsibility lies?”
These issues and risks around susceptibility to hacking were unlikely to delay the role AI would play in primary sector robotics.
Duke saw the technology as leading a complete new industry in New Zealand.
“We can see these machines not only making NZ a more efficient primary producer harvesting consistently high-quality produce but also the start of a high-end, quality robotics industry with great export potential.”
