Think about robotics & automation and the mind typically turns to industrial robots toiling on production lines, but these technologies have applications far beyond the factory. One company doing groundbreaking work in the agricultural sector is Agerris. By Brent Balinski.
Though Australia lags behind other developed nations in terms of robot use in its factories, there are parts of the economy where our local automation has led the world. Self-driving trucks in Rio Tinto’s mines and the world’s first automated port at Brisbane are two reasonably well-known examples – each involving the work of experts at University of Sydney’s Australian Centre For Field Robotics (ACFR).
“Australia’s always punched above its weight when it comes to automation,” says Professor Saleh Sukkarieh. “We’ve always been the country to look at when it comes to automating mining or stevedoring or logistics. The fact we have lots of constraints in this country kind of pushes us down that path and it makes it a great ground for that. There’s some really good companies, really good robotics research institutions, and a desire to see automation.”
Sukkarieh began at the ACFR as a graduate student in 1997, and served as its Director from 2007 until 2018, when he stepped back to take agricultural robotics R&D into the private sector. He still oversees PhDs and research on environmental robotics a day or two a week, but “the other 250%” of his time is spent as CEO of Agerris.
The company’s main product is Digital Farmhand, an autonomous, solar-electric four-wheel drive cart designed for smallholder farms. It can currently cover between three and five hectares a day while building maps, gathering intelligence on soil and crops, and weeding without chemicals. At only 300kg or so, it is far gentler on the soil than a tractor, and with none of the greenhouse emissions.
The Professor of Robotics and Intelligent Systems – Agerris is a contraction based on ‘ager’, Latin for ‘field’, and Robotics and Intelligent Systems, RIS – was first drawn to agriculture through research on drone identification of woody weeds. Years of smart farm-themed work supported by organisations including Meat & Livestock Australia, Horticulture Innovation Australia and others was spun out in April 2019, with the official launch of the agtech business following a $6.5m seed round.
Out on the farms (the current focus is vegetable growers), the owners quickly overcome the shock of a new machine that is both smaller than they’re used to and non-diesel-powered, as well as being able to go eight to 10 hours up and down rows on a single charge.
“That firstly takes them by surprise,” says Sukkarieh. “And then they can see the benefit of the fact that all you do is just plug this in and it has one-tenth the moving parts that a tractor would have because there are much less moving parts, no oil, fuel, lubrication, nothing like that”.
The company’s machines cannot yet harvest anything physical, but their sensors are constantly collecting data from crops and soil, with this processed into maps and analytics down to the level of individual plants.
“Initially we were just focussing on the robot weeding,” Sukkarieh adds. “But then when you start to provide them with the crop analytics, they could see significant benefits that came out of that process.”
The data can be used to chart the rate of growth for plants, for yield prediction, and to detect diseases before they spread.
According to Sukkerieh, a long-time researcher and first-time entrepreneur, things are progressing better than expected. Instead of building a couple of robots for trials, the team now has 14 Digital Farmhand commercial prototypes out in the field.
“We’ve learnt a lot about how to manufacture efficiently. We’ve learnt a lot about operationalising and putting it on-farm and making that work.”
Scaling up
Initial units were designed and built by the Agerris team, with structural parts made overseas, with Agerris then turning to Chess Engineering as its manufacturing partner. The ag-tech start-up creates its mechanical, mechatronics and software designs in-house. As the company started to scale up, it handed these over to Chess to make the product more robust and help with design for manufacturing.
“They also integrated some of the motor-drive units,” adds Sukkarieh. “And our role then was to kind of integrate the rest of the electronics and sensors and get the bot working.”
Getting to the manufacturing stage has provided lessons for the team.
“When you start off, you do a lot of customisation and you’re trying to build a product that would be useful across some different use cases, and as time goes on, you learn how to solidify on some of those thoughts and focus on locking in certain designs,” explains Sukkarieh. “You can see that if things work out and we scale further, that you end up with a bit more of a common design architecture, which is easier to build, easier to manufacture, and I think that’s probably where the future focus is. So how do you do this in a way that is easy to build and hence cheaper to build? Because obviously those costs go on then to the farmers.”
The company has recently been engaged in a $5m capital raise to move into higher-volume production, reach new markets, and continue R&D. Part of the R&D program involves working vertically into agriculture, including plans to harvest crops as well as understand them. Picking presents a two-fold problem, according to Sukkarieh: environmental, and technological.
Robots have been in factories since the 1960s, where they work in structured environments, generally unable to adapt to changed circumstances and simply going through a pre-programmed set of instructions. A farm is not so structured: row widths aren’t standardised; every farm is different; and outdoor conditions (light, temperature and other environmental factors) are always changing.
On the technology side, produce is tricky to get a handle on for a robot: it’s delicate, can be obscured by branches and leaves, and the way humans grasp is more impressive than we take for granted, says Sukkarieh.
“The perception system of a human being is quite phenomenal, and the internal models that we have in our brain are phenomenal: when we try and predict where the fruit is, what angle we should pick it from, how we should pick it, how soft do we grab it, etc. So all those things into the technical space are important to capture.”
What everything’s tending towards
Farm labour is short at the best times, and pandemic-related border closures have worsened this. In October, NSW Farmers President James Jackson said: “We’re getting into a pretty tough spot with a lot of crops. We need up to 16,000 farm workers and we need them now.”
COVID-19 “hasn’t been rosy for anyone” according to Sukkarieh, who says that while interest in robotic farming has increased, his company has itself had lower access to labour and components sourced through its supply chain.
For agriculture, it is arguably a sector in need of reinvention. While Australia exports 70% of what it grows, and its farmers are sometimes referred to as among the world’s best in terms of efficiency, the average age of a farmer is 58.
Technology is one possible way to attract younger talent into the sector and the regions where it’s needed.
In the same way that modern manufacturing – high-skilled, high-tech, highly-removed from the dirty, giant factories of yesteryear – might be a more appealing possibility for a person planning their career, there are similarities with this new farming paradigm.
Agerris is focussed on the lighter, nimbler end of the industry – solar electric-powered and based on a vision of people working alongside teams of robots, each perhaps specialising on a different task: tilling, cultivation, weeding, and so on. Rather than being based on giant farms, giant machinery and physical strain, it might be based on giant volumes of data, and the exercising of creativity to solve problems based on this.
Time will tell how the sector adapts to its challenges, and what its future workforce looks like.
Sukkarieh sees the nascent agricultural robotics genre being helped along by smaller, more powerful, electric vehicles with better endurance, with bespoke systems easier to produce through improvements in 3D printing, and with more powerful processing meaning each robot is smarter and more capable on-farm.
“You can imagine what everything’s tending towards, which is a smarter, more nimble platform that can do precise activities on-farm and hence reducing the environmental footprint,” he says. “The only thing I think that hasn’t changed is the amount of capital that’s available in the industry to be able to kickstart and push through automation.
“And if that was available in this country, there would be no reason why you wouldn’t be able to see, in a decade’s time, fully autonomous farming activity, from planting through to harvesting in certain commodities.”