 Good afternoon everyone. It's a pleasure to be here today and thank you for the organizers for allowing me to come and give a talk here Currently CEO of a startup. You guys started in April last year Prior to all that I was an academic for about 20 years working with large industries around Australia mainly mining aviation and stevedoring focusing on not just autonomous robots, but also autonomous operations and In parallel to all that about the last 15 years. I've received Just over 30 million dollars worth of R&D funding from various RDCs to try and look at what does it mean to build on-farm robots I have field robotic tools things that can operate 24-7 in different settings and can do certain tasks on farm And they really around those that last 15 years was really around exploration of those different types of activities So the way I've broken up this talk is I'm just going to talk about Bots that we built that we thought might be useful for the top 20 percent bots that we built that we thought might be Useful for everyone else and then I want to touch a little bit on on thinking about what does it mean to go from? Just robots to operations and and the reason why I wanted to touch on that is because as Growers as farmers what you want to be thinking about is what does the operation look like in five ten years time? To dictate about what you want to do now with the robotic solutions as opposed to the other way around So I'm just going to touch a little bit on that and really just by giving some examples from other industries that we can learn from in Australia The first I'm hoping that there's no volume on this bit the first bot that I'm going to talk about is Ripper It was built for the horticulture industry solar electric 24-hour operation on a good day like today Runs about 12 hours of batteries, but it runs the solar energy is just sufficient to kind of move the bottle long Can do real-time AI detect individual plants and start to look things such as crop growth monitoring As well as plant health as it's going through Because you can take the plant you can also start to do things like your individual plant yield estimation And that becomes a significant tool If you can take the plant you can also detect other things such as weeds in this particular case in a bit We had a big push in terms of looking at non chemical weeding and this is just one of the tools We developed a whole bunch of different mechanical tools laser tools steaming tools to try and look at how you might remove weeds without any chemical use at all and We also looked at other aspects. So for example, you'll see here the directed spraying So this is the targeting every single plant and based on the size of the plant measuring the right amount of fluid You want to disperse on that individual plant as you're going through And we also looked at water monitoring and bring up a look at water conductivity So being able to measure the kind of process near a plant The big thing that I want you to kind of actually I'll show you one more after we go through This is it really just kind of mapping out the what's going on under source. So it's constantly every few meters It's giving you a sample The other thing the farmers were always asking us about was you know foreign object detection And whether we could actually remove foreign objects So we just tagged a Ryobi vacuum cleaner we bolted on underneath the bot And then we just looked at whether or not you could just detect foreign objects They wanted us to suck up frogs and mulch them, but the University Ethics Department wasn't going to allow any of that So it was just purely just post-it notes for now But what I want you to imagine is the capabilities now of what we're getting to which is this 24-7 bot that cannot only just move up and down maybe do some weeding tasks or something like that But you're also detecting individual plants. You're sampling at the soil. You're acting on decisions So a lot of stuff that we've been talking about about information to decision-making It's starting to get close and closer where that's where that temporal space is becoming very very short and it's happening in real time On the bot on the farm and hence the relationship between the farm and the bot is more about when you see certain things This is what I want you to do and starting to increase that autonomy on farm We looked at how we could spread the capability that robotics across the different industries because At the end of the day, you don't want to build a robot for a certain farm and then build a different rubber Different from you try and look at scale and from that scale trying to reduce the cost that it makes it easier for the farmers to adopt This is another example of the bot going up and down tree rows little white box that you see up the middle There is detecting individual apples doing crop yield information as it's going through We've also looked at detecting flowers as well So you can start to look at Assessments between flower and apple count in this case here. It's got a little viper You know, which is just firing away fluid for us as roboticist. We were just mucking around We just thought we'll just try this little fire every apple that looks like fun But the growers were coming on there saying well, that's great If you can do things like that then it's directed pest assessments and being able to direct any certain amounts of spray at particular parts of the tree So it's things like that that we were trying to look at as we as we spread through One of the recent projects that we're looking at is with the grains industry and that's looking at green on green mechanical weeding And again using the bot. We just kind of cut it out a little bit There's little pod that you see on the left there sits underneath. So this pod here sits underneath the bot there and basically it's A pod that splits apart the grain Plants and then underneath there. There's a little camera system looking down and it's detecting individual weeds And if it detects a weeds, it would just mechanically chip away or grind away at that particular weed Okay, so you'll see here The bots moving along and here it's doing real-time weed detection as it's going through Detecting we've been a turnip grass as well as very turnip weed as well as rye grass So it's just going through that process and detecting each one of those and this can operate it You know if you're looking at that speed we can operate probably twice as fast as now But again, you're looking at 24-7 capability as you're going through And here you'll see this little bottom picture here is just showing you when it comes along on just kind of little grinder That just chips away at that weed and just when it sees a weed it just comes along and it will just grind away That's just one of the tools. We also looked at we also looked at laser weeding Which which offers some promise here The nice thing about laser weeding is that you can actually direct it to the individual weed and add a particular location If you get the frequency of the laser beam, right? You actually boil the cells on the inside of the plant So you're not actually Bursting the plants at the flames or anything like that. It's really kind of like an internal self-destruction process So if you can just kind of get it's almost like microwaving it But if you can get it right you can actually boil the the cells and turn or just destruct the cells internally Something completely different. We're looking at other bots. So this was for the grazing livestock industry I'll pause just for a second before I play it. This one here was a swag bot. We called it But it's a four-wheel drive four-wheel steer bot. It has a little pivot point on the middle So allows it to go over logs and rocks and into ditches and so forth When we first put this into the industry the grower was like, oh great We want you to herd the animals and so we did the classic thing Which is drive the bot into the animals and watch all the animals run away and nothing worked at the end of it all But what we then realized was that as the robot spent more and more time on the farm and also with the farmer The animals got used to the bot Okay, and so then all we had to do was kind of put this little trailer on the back with a bit of hay And just move wherever we wanted to go and the animals would just follow behind it So it was a bit of a pied piper approach But what the what the growers were interested in is if the bot can measure things like pasture quality and pasture biomass Quality is still a very hard element. We can't do that one yet But pasture biomass then you move the animals to whatever location you want and then you can you know have them grazing But at the same time what you can do is that this is in a in a dairy At the same time what you can do is you got the laser units that are on the bot So as you're moving around the animals, you're also detecting the animals themselves And you can actually apply machine learning techniques On the data and start to see the walking gate function of the individual animal And so what that happens what that means is that if their walking function their gate actually changes week on week Then that's an alert for the grower So you're moving the bots around so you're moving the bot around you're measuring pasture You're getting the animals to move with you and you're also detecting their behavior at the at the same time And and and the other third area that the farmers were interested in was weed detection This is the bot running around in real time detecting serrated tussock As it's going through it's navigating and looking at obstacles moving around obstacles Detecting where the weeds are and it comes along and positions itself on the top of each weed And it drops down this little weed spray and it just sprays and in our case We just sprayed some dye just to mark out what was going on But it would just come along and just spray on the weed And this is where you can start to see the cape You know the reason why we went the four-wheel drive the four-wheel steer just to get that precision And being able to detect individual plants as we're going through Okay So that's on on on some of the on on some of the mid-size most of these bots are probably Let's say a four square meter or two meters by two meters That's the kind of rough size that you've got And the cost of these bots can get quite high because the farmers that are interested in the bots are not just interested in the Labor saving aspects, but also the crop intelligence or the animal intelligence comes in So the sensing capabilities the algorithms that come through all kind of phrase it But we also looked at what does it mean to kind of look at the other 80 percent Of growers out there and we kind of this was about six years ago This is like almost like an inverted segue. There's some smarts in the wheels That allows us to kind of just bolt together quite quickly within about 15 minutes You know the wheels that bar that you see in the middle can change for different row widths And what it means is that you can kind of roll it out into the back of a vehicle Go from one paddock to another paddock put it together and you start collecting some data You'll see soon the the bot here moving down. It's got the sensor pad down the pond The rocking motion that you see there is part of the reason why we look at stability just in terms of two Two wheels but when it comes to the algorithms the algorithms can detect the motion of that Sensor platform so it stitches it all together as if you were flying a drone But what this gives us is a six hour battery life So we operate for a lot longer We can go for a lot further and the resolution can Is much higher because it actually goes down closer to the crop And and so you can use start using low cost sensors as well We've done a number of different trials with growers all around the country different local land services Demonstrating the capability of it if you if you look out the front here if you can just see it There's this little thing that's sticking out there and on the top there. That's a selfie stick It's gimble. I said it's just an off-the-shelf selfie stick We put a smartphone on it and then we use a smartphone because it's got the best computational capability and camera sensor That you could buy for a few hundred bucks So all you're doing is you're tapping into it now and you're extracting out the information And you're running the algorithms in real time and you kind of see this kind of thing process happening So that's a selfie stick with a smartphone on it and it's gimble. I so it keeps everything nice and flat It's just doing yield estimates. We can also do things like Seeding and spraying and the phone talks to the rest of the the system as you're going through Part of the drive around building this tech for those small growers There's also what we can do around the digital divide because it's you know from sydney and so it's great You've got all this powerhouse of knowledge and geeks who want to kind of run AI algorithms and that But what do you do about the farmers who are out in the middle rural areas? How do you support them? Last thing you want is collecting all this data Shipping a gigabyte of data across the sydney us analyzing and fritting back So a lot of our drivers also been about how do we build the community around there? We've had a program now and we just got funded again by the new south wiles government We're actually rolling out the robots across 20 rural schools in new south wiles That's quite an important element because we actually leave the robot with the school for 10 weeks for a whole term And we're given material that's connected to the curriculum the strain curriculum And the kids learn about technology in ag but in the context of the robot and they get to program the robot They learn everything from GPS right at the beginning all the way through to machine learning right at the end And how it gets used on the on the farm the whole idea is to kind of encourage the kids Ag is great ag techs coming stay on farm etc But also to kind of build up that community around the farmer and around that digital technology process And it's an interesting area because when you talk about robotics and the future of robotics on farm One of the things that kind of battles us at the moment now is are we building the robot for the 40 50 60 year old farmer Or are we building the robot for the 20 year old farm because they're completely different user interfaces For the 20 year old farmer the ones coming across in five years I'm just going to throw a joystick at him and I'll walk away and they'll do everything from there And when you when you're looking at the older growers You actually have to sit down you have to think about the user interfaces the methods the process of whatever and that You know makes things a lot more complex And that's the kind of scenario that we're working at with robotics is is kind of dealing with the current situation now But also how what do we need to do about the next generation? We also got some funding from the government to actually take this spot across to indonesia Samoa and Fiji and I'll just stop it there for a second and you might think why would you take it to the pacific islands, right? Exactly the same problems the growers that we have here Dad's going off into the city because they need dual incomes left with mum at home working on the farm Kids don't want to work on the farm They've gone off the city try to get a tertiary education moved on from there And so you're getting lots of cooperatives mainly women driven cooperatives that are coming together To try and deal with the issues on farm. They can't deal with agronomy services that easily no one goes out onto their farms and helps them any of them Usually generally the chemicals that they use are cheap They have to get them cheap because of the economics that they have that come from china It's all in chinese. They can't read it and they can't deal with the pests and diseases that they've got they don't understand It so being able to come in with a solution that can Provide that crop intelligence as well as that labor saving elements for these farmers is actually a big thing And so we we we did some trials out across those three islands This is you don't need all these people they you know, they were just standing Interesting story when we went to samoa. This was this is about 300 kilogram bot It was in a crate We got there and it was still in the crate in the corner And we thought how'd they get it off the back of the the ute onto there And we needed it moved and we're like we could looking for a forklift and we couldn't find anywhere And the guy the hotel just called up four of his mates and I just Picked them big small ones right they just picked up this thing and moved it across And so what they thought was you know, you got to worry about this thing But this is just hinged down as you go through the back of the ute So it's just a hinge in it goes hitch. Sorry, and it just goes through And then it's the same kind of concept You've got this bot that's going through cheap sensors looking down providing crop yield information and and Sensory information as well as being able to do certain tasks on farm And we modified the same bot to look at tree crops So in this particular example, this is the digital farmhand It's meant to be the low cost version just trying to it's almost like a two-wheeled electric tractor But with dolly wheels on it allows us to put all the sensors on board and and off we go from there The capability of this particular bot was about four to five hours of battery life But what it's doing yet It's don't look at the spraying. It's just us having a bit of fun But they're independent actuated nozzles and what it's doing is detecting the yield of fruit on the tree And based on the yield estimating which of those nozzles should move away in order to be able to target and spray at that particular point So it's it's really putting Robotics and intelligence together by doing it in real time. Okay, and that's really what the what the what the focus is So in april last year A lot of that technology got spun out We we got three investors that come through gave us one of the largest ag tech seed funds about eight million dollars To kind of really put about 12 engineers Together and start to take out the technology and start to commercialize and work directly with the growers in that way The first thing that we've got out is the digital farmhand. It's it's that lower cost entry point It's it's looks it's a small bot. It's about Uh, the width can go about 1.8 meters the fitter is rose even two meters in some cases But it's adjustable as you can see there and it's about one and a half meters long So we can adjust there but in these red boxes here is where all the energy drive and the battery systems are And the black box that you've got up the front is where all the smarts are and what that means is that we can come Along and configure the bot for different types of arrangements and that just makes makes it more adaptable and usable across different size farms It's quite capable. It can move at about Six kilometers per hour at top speed. We can adjust the speed or it's a torque speed relationship It has a seven to ten hour battery life and we've total lane cruiser with it. So it's quite a capable bot It's even for its small size and that gives you a perspective of the kind of energy that we've got now This is just a video kind of showing the the latest results that came out of the the bot So this is on one of the farms out in greater, sydney Ignore the noise. So this is just us adjusting it and we can just have a different width It's doing broccoli estimation One of the things is trying to measure different size for selective harvesting And that's one of the kind of areas that we're looking at here It's doing real-time weed detection amongst in a carrot farm on the bottom left and it also does row following So it's not just gps guided. It's following rows. It's doing obstacle detection Noticing when there is an end of a row being able to do a turn autonomously and coming across back into the Into the row again So we're trialing this spot We've got a number of early adopter growers both in victoria new south wales And we're going to be positioning ourselves with those growers and and it's going to be on farm and really working towards that commercial strategy of It's not just weeding and spraying but looking at the digital agronomy element as well So how do we do you know the individual plant detection? What does that mean for the farmer and what does that mean back in terms of the operation? So the last thing I wanted to touch on was just how we move from robotics to operations and and again There's a lot of hype around the robotics Which is good, right? They're coming the text coming It's it's going to get better and better and it's going to get cheaper and cheaper And that's one of the things as farmers you've just got to it just accept the fact that it's coming along So the question from an operational perspective is is how you use it within the context of operations I want to bring out three examples of of projects that i've worked on But also they're Australian examples that we can learn from if you go to port brisbane port melbourne Port sydney you'll see that there are autonomous straddle carriers moving around autonomously 24 seven We developed all that tech from You know about mining and what's going on in in Perth the whole mine operations that are going out in the pilbara Again, we developed a tech for rio tinto in that space And recently you would have heard about the london to sydney and new york to sydney direct arts And we developed a lot of flight planning algorithms for quantus in that space But what you see in that trend is this focus on what does the future state look like And hence what is the technology that and we need to get to at that point So as an example if I was a horticulture grower and I knew this technology was coming What does an autonomous farm look like in five years time? What state will that look like and maybe not and not everything will be automated But what are the things that will be automated? What does that mean in the context of my operations becomes another question And all of these guys have got examples of things that the agriculture industry can learn from and they're already there already set in place from standards whs How do you run autonomous operations and so forth? And i'm just going to bore you with the it shouldn't be the last slide but I will try and Yeah, I just want you to kind of see it in this context This is a this is an activity of a mine operation And this is the number of years across here and this is how much uncertainty there is And in the light blue what you have there is what the current operations are so the higher the value the more uncertain Everything is And what you see in the dark blue is the current operations So in other words, they've got more information about what's going on on their farm And because they have more information they have better decision making But what's more important is in the process of automating some of these things There are many on mine tasks that have been completely removed from the operation So in that context when you're looking at the situation about your farm You're not just you're not just looking at and saying oh, these are the tasks that I can automate But if you look at it from a longer-term horizon, there are certain tasks that if you automate it There are other tasks that you don't need to do anymore Okay, and it's that thinking that looking ahead of the next five ten years It's going to help not only yourself in terms of understanding how to bring in the technology But it's also going to help all the ag tech providers understand what the key points are that need to be brought across So that's the presentation. Thank you