BCG acknowledges the traditional custodians of country throughout Australia and their connections to land, sea and community. We pay our respects to their elders, past and present.
Welcome back to Shared Solutions by BCG. I'm Janine Batters in the fourth episode of our series On the Horizon: Farm Tech, sponsored by GE silos, I'm chatting to Bailey Kilpatrick from SwarmFarm about their autonomous farming technology. Welcome, Bailey.
G'day, Janine. It's good to be here.
So good to have you today. Now, first I thought I would start. Can you just give me a little bit of a history about SwarmFarm , please?
SwarmFarm originated on a family property in central Queensland, about 40 minutes south of Emerald. And it's designed with the agronomy of farming in mind. SwarmFarm started on farm, is designed by farmers for farmers and all Australian owned and made.
Okay, tell me a little bit about your role.
So I work in the Central Queensland region as a portfolio manager. And what that means is basically that the growers that are running the machines today, I look after the operations side of that, and then some of the sales for the new ones coming through.
So can you tell me SwarmFarm makes SwarmBots ? What are SwarmBots ?
Yeah. So a SwarmBot is a robot that we build. It's a it's an autonomous platform that is designed with a farmer in mind to then put an attachment on the back of it to then. Yeah, go out to the field and do the job.
So how do they actually work?
There's a few parts to them. The main three components or the main three parts of a swarm bot. It's a diesel engine, so it's a diesel powered machine, four wheel drive. And they run only inside a mapped area. So you've got to have a geocache like a mapped area for them to then traverse and drive within.
And so if I wanted to get a SwarmBot, but what would be the first step?
So the first step is looking at your farm and going, righto, control traffic three metre centers and I'm 24m. If I want a robot, then we provide the orange and silver bit. And what I mean by that is the the technology for the dare I say it, tractor. Andrew is going to kill me. The the robot is what we provide. So the first step is looking at your your farm and then deciding what configuration you need, and then pairing that with an attachment that goes on the back of our robot,
that then from start to finish we'll support it. And then there's a layer of training that's involved and mapping your farm or bringing maps in so the robot knows what it can and can't do.
Okay, so knowing what you want your robot to do, knowing what sort of tramlines you want it on, and then educating yourself on how to use it. So are there different models?
There's only one model. So the history of Swarm, circling back to that, it started the very first one was a like a John Deere Gator that did a project with us on to do an autonomous light vehicle. From there it grew to a three wheel electric powered machine. And then after the three wheeled one, the the four wheel design came out, which is SwarmBot five, what we call a SwarmBot five now. So that's the one model, so to say, that we build.
75 horsepower hydraulic drive and the variation comes in your axle sizes and axle heights to suit either two, 3 or 4 metre centres. If you're in an orchard, you want a low to the ground two metre wide machine perfect for that. If you're in Broadacre, which has sort of been our big uptake market, that's sort of been the lowest hanging fruit. What we've done most in is broadacre. Generally, three metre centres is pretty common and what we'd call
a high chassis. So that's the next step up in axle size just to get a bit of crop clearance.
Okay, so have you thought about getting any bigger?
Not in terms of weight and size bigger for different applications. So what we need at the moment is powder ground for ground, engaging gear. Our our market has been in tow behind sprayers. They don't take a huge amount of horsepower to pull. It means we can keep our weight really light for a 24 metre boom. For example, will be eight ton loaded so you can get on the ground nice and early. We don't want to go bigger
because that's going away from where SwarmFarm began. The idea with the Swarm is you've got lots of small, lots of lightweight machines doing the job that you'd think of one big one doing. So the reason why we aren't going bigger is because we believe that you can do
the same job with less horsepower and less weight. There are limitations, of course, once you get into planting or bigger mowing or raking, that does require some more design, but we've got a full mechanical design team and everything's done in-house from the software development, the chassis design, R&;D, everything's built and designed on farm, so it gets tested here and then we break it, fix it and then take it, take it to market after that. That's why we're not big.
It's all made in Australia. That that is a really cool point. So I'm just thinking in terms of you're talking about how it's being used in broadacre the optical sprayers. Are you saying, would you use a lot of these
SwarmBots instead of a big self-propelled? Or are you saying that you would have a self-propelled for when you really need to get over it nice and quick before a rain, and have the little bots going when you don't necessarily need to be done straight away, but you want to get the job done.
That's a really good question. So Andrew's brainchild, the concept was to have lots of lightweight machines, hence the name Swarm that's grown into what we have now. So our largest boom is 27m wide and we tow a payload of just over 3000l. So 3000 litre chem tank with a freshwater flush tank is our. That's our largest setup at the moment. We have a maximum speed of ten kilometres an hour and we generally work if you have a 24 meter arrangement, 20 hectares per hour is what
you'll benchmark on. So if you're looking at a SwarmBot as a replacement for your sprayer, it's probably not what we'd specialise in. It's more to pair with your existing spray setup, so we don't want to replace your Broadacre sprayer. That's not really what we're about. We're more using this as a tool, as an autonomy tool and a maintenance tool to go across your paddock more frequently with lower
water rates, high chemical rate with a spot sprayer. And that's sort of where it's fit into the industry really well, because you don't need to go out there and race into your paddock and do it in a mad rush because the rain's coming, because you've already been across it three times with your SwarmBot . And that's where I don't think we'll replace a self-propelled. They've certainly got a seat in the industry and wouldn't say that we're replacing them,
but it's sort of working with them to go. You can put the hours on a SwarmBot , and it's a better return on investment than, say, putting 10,000 hours on your self-propelled machine.
How many hours can they take these SwarmBots before you have to recycle them?
Our end of life is around 15,000 hours. And what I mean by an end of life is a full rebuild. So at the moment we've got machines running with 10,000 hours plus and they're absolutely fine. We are new to the industry, so we're still learning and growing. Um, but that's that's our lifespan is 15.
So I was thinking when you were talking about having it running and having the self-propelled and the SwarmBot that perhaps you could just let it do the jobs that weren't high priority. But I was also thinking about perhaps you're getting a better kill because you're going to go over that paddock when maybe you wouldn't have otherwise because you don't have the time.
Yeah, absolutely. It's coming into an agronomy decision more than a time based decision. So they're not a self-sufficient machine. I suppose it's a question that we get a lot. Maybe a misconception that we have is it should do everything. Why have we got to look after it so much? It's another little worker. Yes. Once you do the mapping and planning and you have your farm and your your maps laid out, you do the rest from your phone. So you operate it on an Apple iOS device and
that's your that's your interface. So that's your tractor screen is your phone. You walk around with your tractor screen in your pocket and you can check on it. It'll send you notifications of when it needs filling up fuel,
when it needs chemical, it'll send you. If it's a sprayer with a weather station fitted, it'll send you notifications for the weather parameters that you've set if it gets close to the boundaries of those parameters, if it leaves the parameters, if it stops for an obstacle, it's definitely not a set and forget machine. They do take some running, but it's freeing up, I suppose, one of your workers to be fencing or checking troughs while they can look
at their phone and go yep, Bob's still spraying. He's got about 50 hectares to go. And that's that's sort of where they come into their own. So you can set it off 6:00 at the night, it'll start spraying and then through the night if it goes outside the spray boundaries that you've set, it'll pull up. Go to sleep. We've got a sleep passive state where it'll just shut the engine off and keep the GPS running, keep the weather station running. That sort of technology will stay on.
And then once it's back inside its boundaries, that's perfect for spraying. It'll start up and it might be 2:00 in the morning, but it'll continue on its way. And when you wake up out of bed and you're like, oh, how good is that? It's done.
That is very good. I like that you've applied that to sleep too, because that's so important for farmers. So you're talking about you needed somebody to operate it all the time. Can you talk me through the process of what that looks like on your. Like how do you just get it set up on your iPad because it only works on Apple. That's right.
Yeah. Yeah that is correct. So Apple's the um the only device that we're integrating with at the moment, from the initial stages of getting your SwarmBot , you'll, you'll integrate either your pre-existing maps from your John Deere Ops center or Trimble maps. We have a farm manager website
that is needed to operate a SwarmBot . So we've got a dedicated software team that's built a platform where you can bring your maps to farm manager or grab a GPS drive around in the buggy, map all your obstacles, map all your paddocks, get it set up so you know exactly where it's going to spray, where it's going to start and finish and fill points and map all your internal roads, etc. where you want to send it. But yeah, it does take some running.
It is good though, that you can use the maps that you've already got because they take a lot of time, so it's nice to just be able to transfer those across in regards to just maps in general. Bailey. If it's taking maps, if it's making an EM map or something like that, can you integrate those with your other software?
Unfortunately, we can't just yet. So specifically for an EM unit, that's a another supplier that's attaching an EM unit to one of our SwarmBots and then using the SwarmBot to continue on the path. So instead of driving up and back in the ute , the SwarmBot can do 7 to $10 an hour, which is ideal for EM mapping, and it'll chug along there and just go up and back your AB lines, and then from there it'll send that data to that suppliers.
So you said that they can run all night. What about filling them up? How long do they need to go between fuel fills.
The the fuel usage from the engines about eight liters per hour. So they're very fuel efficient depending on the payload. That's sort of the weed bank that you're spraying in the field. So ideally you can start it and it'll run all night or all day and you just fill it up once a day. That's the idea, especially with a spot spray that's viable. Um, if you're trying to blanket spray with them because you can also send them to PWM and blanket spray. That'll require filling up more frequently.
The perfect situation is having a dock station, um, which we're in commercial testing at the moment. The scenario there is you map your refill, it'll spray until it gets to the low payload that you set via your app, and then it'll go, no worries, I'm at 50l in the tank. I'll fold myself up, drive down the road, fill up, go back to my last spot where I sprayed and continue on again.
And by road you mean road in the paddock, don't you?
Road in the paddock. No public roads can be mapped.
Okay. That's good. I think I'd be a bit freaked out, actually, if I saw one coming along the road. So can it do that with chemical as well? Is that also in testing.
So that is that's only for chemical.
Oh so that's not for fuel.
No we're not doing fuel. So they still require checking every 24 hours. It's still a machine. It's got hydraulics. It's got a diesel engine. It's got a steel chassis. So it still takes a daily check. Walk around, check the sensors. They're really important. We've got our obstacle detection. That is, um. Yeah. Is vital for keeping it and everyone else safe. So daily checks are still a must.
I like that, I think that's really clever because I just imagine that. Yeah, I just want it to run all the time. But then I'd probably yeah, wouldn't probably check it. So that's a really good point. So leading on from that, what about servicing. Can you do you have to take it somewhere special. How hard is it to service it and get parts?
Everything has been designed to be very user friendly in terms of servicing and repairs and maintenance. It's built with right to service in mind from our point of view. We warrant hardware everything other than oil and tyres for three years. The replacement of parts. We've got regional hubs set up around Australia, so if you've got any issues, we've got a full time support team and full time,
I mean, seven days a week within specific hours. It's not 24 hours just yet, but yeah, we've got a full support team that is at call and ready to help.
How hard is it to get parts once you realize you need something? What's the turnaround.
For major parts? It depends. If it's a proximity sensor, we can have it there within the day. Depending on where you are from the closest regional hub, you also get a little toolkit with the major components that are easy to swap out. Um, proximity sensors, fuses, that sort of stuff that comes with the robot. So a lot of that's sort of you can fix yourself. It's growers right to service. If it's a major component, like a wheel motor or a big failure, then 48 hours is
our target for a full downtime. Back up running.
GE. Silos have been leading the way with their custom built grain pellet and fertilizer silos since 1978. Built using Australian made steel and offering a large range of high quality standard and optional features, the team at GE silos can work with you to ensure you find the right silo to suit your needs. Contact. GE silos today or visit GE silos. Com.au for more information GE silos the right storage solution.
What about being a robot? What about if it's something technical?
With our connectivity, we can log into the robot and see it remotely. So imagine you've got your phone walking around farm, anywhere you can get phone service. That's what the robot will be able to operate and transmit messages to you as long as you've got the machine in phone service, we can log in and check it. If it is a no service area, it's it's not going to be detrimental to the machine. It's still got its pre-planned path and it'll still operate. The only time there'll be an
issue is if it stops from an obstacle. It won't be able to send you a notification because it doesn't have any messaging capability, doesn't have service.
Do you mean that then for it to work to its best, you're going to need good connectivity.
Yes, yeah I do. So on your farm you may have Optus as your best service provider. Maybe Telstra. They take a SIM card in a little, um, router and that's how we can see them remotely.
So in areas that don't have that can use another internet provider, something like that.
Yeah, 100%. We do have quite a few machines running with the Starlinks in a hard case, we can mount them on the robot, and they often give better connectivity than you've got at your house, which is pretty funny. You can be out in the middle of the paddock and load your messages faster than at home sometimes.
That could solve a lot of problems, I think. Bailey. So you said that they travel about ten K's per hour? I think you said. And they can cover 20 hectares per hour. Is that right? Yeah, yeah. So what in terms of cost, how much are they compared to something that's going to do it a lot faster.
For the machine? What we build and supply is just over 300,000. So 305 gets you a robot, your first year subscription and full service and warranty from there, once you've passed the three years, $12,000 per year for your subscription base, with that, you still need to put an attachment on top so you've got a couple of options with your chassis and axle sizes. Other than that, it's just the one make and model. So 300,000 buys you a robot, and then you put the attachment on top of that.
And can you just use any kind of attachment or do they have to be SwarmBot attachments?
They've got to be SwarmBot ready. So we've worked very closely with attachment suppliers. From the beginning we recognized that we can't do everything. We don't want to do everything. We don't want to build sprayers or slashers or planters because that's not what we're targeting. We want to build a platform where then others can integrate with us and get into the industry that way, and better the industry that way. So we've partnered with quite a few attachment suppliers around Australia.
And they are that's their bread and butter. That's what they do every day. So we figured why not, you know, work with the best in the best.
That makes sense. So what what kind of attachments can you attach to it at the moment?
At the moment we can do spot sprayers are our most common attachment and that's been the the lowest hanging fruit, the best entry to market you can partner and pair an attachment on for 250 -$300,000. So you're looking at half a million to get a full setup. And that'll that's ready to go s pot spraying. That's been our main attachment. We've got some turf mowers running in New South Wales and then other slashers running in Queensland in orchards and
one in a vineyard. Spreading is another one. We've got a spreader over in Western Australia.
How big is the spreader?
I think it can hold a ton. So it's not huge. It's an electric driven one, but mainly used for mouse baiting. It's an initial look into the spreading market, and that's where we're wanting to get more and more, is diversifying away from not away from sprays. Still, definitely. That's been a really good market 10-K and hour spot spraying, it's the perfect speed for targeting a weed to weed issue. That's
been the key. But yeah, spraying, slashing, spreading. And we are looking at getting into planting.
Okay. So that's the sort of the near future.
That's hopefully the near future. I think we'll need a bit more horsepower and but we'll see how we go. Yes.
It's really exciting. I was thinking in terms of, hay, just because we do a little bit of hay, on our farm, Bailey, I'd really like to have a SwarmBot that was raking the hay, and then another SwarmBot that was mowing the next paddock, another SwarmBot that was stacking it up, and then one loading it on the truck, and then another SwarmBot back at the shed. Do you think that's something that you might be interested in making?
I feel like we'd be taking, uh, taking too much of your fun jobs away from you.
That workers might not be too happy if we do that. It is, but it is quite intensive. And I see that's where SwarmBot could be helpful, is just because labor is obviously really hard to get. And then there's obviously a lot of work on the farm. So let me know. Let me know when when you've worked that out. So in terms of challenges, what are some of the challenges that you've seen? Well, you.
Mentioned labor there. and I think that's a really interesting point, because when you think of autonomy and where the ag industry is going, we've had in the past where growers think, oh, you're taking a job like, why are you doing this? You're sending a robot out to do what my workers used to do. And I mean, I don't think we've spoken to anyone who's actually bought a machine that has
fired an employee. So that's been one of the challenges that we've worked through and overcome to go, oh, well, it doesn't suit everyone and we don't mind if it doesn't. But I think one of the challenges is labour and overcoming the issue where if you have people knocking at the door, please let us know because we need more people at SwarmFarm.
So what about in terms of like hilly country? Can it handle hills?
Oh, they love hills. Um, they're like Billy little Billy goats. It's amazing what they can climb up. So from a SwarmBot's point of view, we can go up and across some very steep country. It depends what you've got hooked on the back of it, I suppose. We are only 75 horsepower, so it is quite a low horsepower machine. That's probably our biggest limitation in large contour areas. It'll slow down a little bit climbing over the hills, but we haven't. Yeah, we haven't come into a spot that
we can't get through yet. Andrew likes to say it's the little engine that could, because a lot of the time the the tractors that are out on market today, like the largest, is over 800 horsepower. Now you do the same job at a slower speed, which means you need less horsepower. That's all we look at, so you can get across the same country. It might take you a bit longer, but you don't need as high horsepower.
What about water? What happens if there's flooding?
They only know they're bogged when they can't move any further. So that's one thing that our weather station on board doesn't capture. So we don't capture rain. But we do capture humidity via delta T minimum maximum temperatures, wind direction. If it starts raining then we can use third party weather stations. We can integrate with them and grab data from inversion towers. But if it's raining on top of the machine, the only way it knows is if it gets bogged.
That's good to know. So what about crashing? Can they crash into things?
No. They, uh, they can't unless you tell them to. So every machine has an obstacle detection system. It's a lidar based system at the moment, so it relies on light feedback from an obstacle in front of it. We can't differentiate between a tree or a person. Um, so it'll stop for anything that's in front of it. And it knows when it's 24m, when it's unfolded, and it
knows when it's four metres when it's folded up. So that's how we get through, um, gateways and in and around paddocks is it knows its width when it's unfolded and it knows its width when it's folded up. If you're mapping still need to map every obstacle that's within your paddock. So trees, power poles, lakes, dams, they need to be mapped because that's how our planning program works, that are then draws a plan for the machine to
follow around that obstacle. Whereas if you miss it, it'll stop and you'll have to go out there and manually drive it. If you override that obstacle detection system, then it will not mind what's in front of it and rely only on the bumper bar which is on the front of the machine. That's our last line of defense, which we prefer not to use.
Does it happen a lot though?
No, not very often.
Did you train everyone really well?
That's right. The training that we provide is. Well, it has. It has to be not I'm not going to say bulletproof. We've had our moments as every company would have, but we strive to provide the best training we can because we want to see farmers succeed with this technology. And that's part of our package, is ensuring that your farm is SwarmBot ready. So we've got our training material that we'd do over a couple of days is our longest
training material. We've got three training sessions, I think coming up next week like it's yeah, we're pretty busy and trying to keep on top of that side of it.
So got to do the training. How long do you think it would take you like in days or to actually get set up? So if I was to stay today, tomorrow I'm going to get my SwarmBot's going to arrive. Can would I be able to use it in a week?
Yeah, absolutely. You I reckon you'd be pretty tech savvy. You'd just smash it out. Um, so for the from when you first put an order in for a robot, we work closely with you to make sure that you understand what's required for a map, what a SwarmBot needs
to operate. If you want to grab the GPS and go out there and and map a paddock from when you start mapping to when you can run a robot within half an hour is is all it would take once you're proficient at it when you first start, we wouldn't expect you to do it in half an hour, but that's where you can get to.
So how do I get a SwarmBot ? Can I get it from a local machinery dealer?
It sounds a bit funny, but our philosophy is not to go towards a dealership model, so we want to keep it independent and then use local resources, local mechanics, to train them up and be able to one day service the fleet of machines in their area. We have an order form and it comes all through. SwarmFarm is how you order one. So we don't yeah, we don't have dealers. We're staying to an independent service model.
Okay. And speaking about services and dealers and that sort of thing. How many of these are operating throughout Australia.
So we've got nearly 100 running commercially. I think it's 96 running commercially in Australia, and we've got one in the US running commercially.
And where do you see the best fit? Like how many do you see the most of them in one particular area and one particular industry in Australia.
For the.
Area?
I would say New South Wales has been a big growth area they're, right into their controlled traffic and keeping their water use efficiency right down and really targeting resistant weeds because that's essentially the best fit has been in spot spraying technology. Um, using an optic camera sprayer, traveling throughout your paddock, you can really do a plant by plant decision to keep it running as a maintenance tool throughout your field. That's where we've really grown in the ag industry.
Now how reliable are they? How often are they going to break down Bailey?
They are.
Quite reliable. I will be honest that we've got a software and full support team that's available every day of the week. So if you have any dramas, whether it's logging into the app, got a leaking wheel hub, got a robot that won't turn on because your mouse have chewed the wiring harness. The best way to describe it is they're reliable, but we've got a full support network .
That is good to know. What about regulations? Because when you fly a drone, you need a drone license. Now, do you need a license to operate these?
No, so at the moment the industry, especially AG isn't necessarily regulated. It's an AG implement. So the restrictions come when you want to move it from paddock to paddock. So you're not allowed to map a public road network and you're not allowed to operate it on a public road network. So from a user point of view, if it's on your farm's private property, it's registered to your farm, no issues at all. If you want to move it from
paddock to paddock, we've got a dolly system. You can hook it on behind your truck and then tow it down the road. In terms of regulations, we've got an internal certificate network that we're bringing in to make sure that the operators are trained and have appropriate training. Unlike international companies struggling with getting autonomous machines out, Australia is quite free with the regulations, and we're trying to do
everything we can to stay in front of that. So keeping signs at the front gate, entry and exit points. So anyone who enters that farm knows there's an autonomous machine running appropriate training and keeping up with that service and support.
That is good to know that there are some safety things because yeah, I know my robot vacuum cleaner would probably just run over me if it could. I don't think it likes me very much. Was there anything else, Bailey, that you'd like to add?
Definitely. Just reiterating that we build a platform. Think of it like your iPhone. That's the technology that we're building. It's something that then you can load your apps on top of. So you've got a robot, and then with a swarm connect attachment, it can go from a sprayer to a spreader. We're trying to get our time down. So that's that's sort of where we're really wanting to head.
And yeah, working with different industries like you mentioned, hay and raking earlier, that's something that we've never really looked into, but I'm sure into the future. It would be a good fit.
One question I like to ask everybody that I speak with Bailey is, what's the best piece of advice that you've been given, whether it's in life, in footy, farming? Tell me.
I love this question. I think the best piece of advice I've ever been given is bite off more than you can chew and just keep chewing. It's something that I got off my dad and yeah, it's rung true so many times.
BCG drives the prosperity of Australian farmers, communities and landscapes through applied research, innovation and events. To find out more about what we're up to, our team and events near you, visit bcg.org.au