Australian farmer Andrew Bate saw in 2008 what many Canadian prairie farmers are seeing now. His management quality declined after expanding to 20,000 acres and it was time to re-examine things.
“We grew rapidly. Buying more country. Buying bigger machinery. By 2008, we got to the stage we realized we weren’t doing as good a job farming with all this land,” Bate said in a Zoom interview.
The wider the machine, the more its ability to follow undulating ground is compromised. There are always press wheels not working properly, sometimes going too shallow, sometimes too deep. Also, the centre section is extremely heavy to support the weight of the folding wings.
“So we thought about the idea of building small robotic machines,” Bate said. “We could get back to growing better crops and the optimal amount of food on our land.”
Bate had seen how autonomous machines were used by the Australian military and mining industry and did some research. But in 2008, machines such as drones were relatively new, not something bought off the shelf.
Another factor was weed control. Bate was applying 70,000 litres of glyphosate each year and it was only a matter of time before resistance became an issue.
“We could prevent it if we could precisely apply certain crop protection products at certain times, targeted at certain weeds when they’re small,” he said. “Very difficult to do with a conventional tractor setup.
“With the accuracy and agility and repeatability of a robot, we can selectively remove specific weeds from the population. Once we have that in hand, then we can start talking about other killing methods such as heat or electricity or biologicals to target specific weeds.”
Read Also
Sask. ag group wants strychnine back
The Agricultural Producers Association of Saskatchewan has written to the federal government asking for emergency use of strychnine to control gophers
Bate said a robot can be programmed to make six or seven applications in a growing season, each different from the others. As a result, agronomists can custom design a prescription map to gradually eliminate the worst weeds one by one. The next target is always the next flush of bad weeds.
On the flatlands of Australia, Bate can have several of his SwarmFarm robots working in a single field, especially during spraying season. He has about 30 machines at work in commercial-sized fields around Queensland.
“These aren’t on research plots or trials or government subsidies,” he said. “These are all commercial farmers who signed up because they saw the benefits and it’s all been developed in-house.”
SwarmFarm creates all mechanical engineering and digital systems for the autonomous robotic platform. It works with other manufacturers for specific solutions such as seeding, cultivation and spraying.
“Some are row crop, others are broadacre equipment,” Bate said. “We cover every device from low-cost to high-end technology.”
At the high-tech end of the spectrum, SwarmFarm partners with Weed-It in the Netherlands, a company specializing in green-on-brown recognition in fallow land. Another partner is the French Bilberry weed recognition system that recognizes green-on-green for in-crop weed control.
Spot spray technology allows chemical rotation while reducing product. Numerous passes of the fields also ensure that product is applied when the young weeds are most susceptible. The new scheme lets him use fewer residual products and more knock-down products.
“This allows farmers to think more openly about their crop rotations because weed control becomes less of a factor. Now they concentrate on what’s better for the health of their crop and their soil.”
Bate said SwarmFarm sprayers have been used for blanket application on large tracts but the platforms are best suited for spot spraying. He toured Saskatchewan and Alberta on a recent trip to Western Canada to gain perspective on how these landscapes relate to those in Australia.
“I have an Australian customer with 30,000 acres getting by with just two robots, although he is now adding two more. Most of our customers have about 15,000 acres, so that’s not what you call a small farm. The row crop guys are in the high rainfall areas, so they farm maybe 4,000 acres.”
He said most big operators use robots in a complementary role, and still pull out the big sprayer for things like fungicide. But they’re finding more uses for the robot and spending fewer hours on the tractor.
“The big game changer for broadacre farmers will come later this year when we introduce autonomous dock and refill,” Bate said. “Right now we have spray tanks up to 3,000 litres, which gives the robot anywhere from 24 to 48 hours on a tank. Autonomous dock and refill will direct the robots to do their own refills and then put them back in the field to keep working where they left off.”
Bate said the prospect of 600 hp seeding tractors may become obsolete if the SwarmFarm concept is applied to seeding and cultivation. His question: why would a farmer put an operator in a 600 hp tractor when four 150 hp robotic tractors can do the job quicker and without a paycheque?
“Look at the engineering compromises and costs in making a 120-foot planter,” he said. “And it won’t give you a top job of placing seed or conserving moisture.
“Then look at a 40-foot seeder that’s more refined and more specialized and does a better job of placing the seed. I think those big farmers will end up with three or four specialized seeders 30 or 40 feet and maybe set up for different crops and different soil conditions.”
Canadian players such as DOT (Raven Industries) have already advanced in this direction with autonomous systems married to disc planters.
Bate points out that robots aren’t constrained by the same time limits as human operators and this has intriguing possibilities.
“The high-speed disc planters are cool, but in a world of autonomous implements, do we really need to go seeding at 15 kilometres per hour if the seeder works 24-7?” he said.
“I think the concept of big and fast is obsolete. I’d rather go seeding at five kilometres per hour and know that every seed is dropped perfectly where I want it. Robotics let us slow down and do a better job planting those seeds.”
Bate is working with a drill manufacturer to introduce a robotic seeding machine by 2023. His next step is highly precise autonomous application of nitrogen.
Safety is an issue with every company that deals in autonomous machines. Bate says he has been working with 3D shape recognition technology, which works well, but has resulted in unwanted work stoppages. He would like to move faster in developing a recognition system, but supply chain problems have prevented him from getting the needed components.
SwarmFarm is compatible with most management apps on the market.
“We run with just about every other play station out there, except John Deere of course,” he said. “They don’t play nicely with their play station, but everybody else does. And we’re integrating ISO BUS into everything. For guidance, we use Canada’s best. We use Hexagon NovAtel from Calgary.”
Moving autonomous farm equipment from field to field is a big obstacle in many corners of the globe. If it involves adding manual steering, throttle and brakes, then the engineering department has wasted much of the benefit of robotics. But if the robot isn’t in firm control, legal liability might bring any experiment to an expensive and crushing halt.
Bate says it’s not much of an issue in agro-Australia, where public roads are few and far between. Most roads belong to the person or company who owns the farm. In many cases, one target field is adjacent to the next so road travel isn’t an issue. When roading is necessary, the robots travel down a private road, reducing risk of accident. The same scenario applies to autonomous dock and refill.
“When a robot has to travel 20 or 30 kilometres, we have a simple low-tech solution. Every robot comes with a front tow bar so you just hook it to your truck and pull it freewheeling to where you want it.”
