# How SQM Used Autonomous Drones to Increase Iodine Yield in Atacama, Chile

- Event: NestGen ’26
- Published: October 2, 2026
- Length: 28:31
- Watch: https://nestgen.org/library/watch/mYlgeru2lJ8/

> The transcript of a recorded NestGen session, for reading or for use with an AI tool. It comes from the video's
> captions; automatic captions can mishear names and terms.

## About this session

Mining operators are under constant pressure to improve output while keeping teams safer and operations more predictable across vast, difficult terrain. Autonomous drones are becoming part of how digital and operational teams capture site visibility faster and use it to improve decision-making.

In this NestGen session, SQM will share how autonomous drones supported iodine-yield improvement in Atacama, and what mining teams should think through when linking drone operations to measurable production outcomes.

## Transcript

**[0:00]** Hello everyone and welcome to another session at NextGen 2026. Well, this session is very exciting. We've already had a couple of really interesting sessions looking at autonomous drone operations in different mining, you know, scenarios such as open pit copper gold mines and platinum shift mines. Uh, this is also a mining environment, but the problem we're looking at today is quite different. Uh, today we're going straight to Chile to SQM's operation in the Atacama Desert where the team is using autonomous drones to monitor irrigation across a huge operating area and ultimately connecting all of that information to iodine production. And what makes this session particularly interesting is the combination of scale, the operational problem being solved, and the connection between drone data and the actual production outcome. So, really glad uh that, you know, we're getting a chance to host this session, and I'm sure the audience uh is going to derive a lot of value

**[1:05]** from it and gain a lot of insights. Without any further ado, I'd like to introduce Rodrigo Toler. Uh, Rodrigo uh is the digital transformation lead at SQM uh and the person behind the company's autonomous irrigation monitoring program in the Atacama Desert as well. Rodrigo, welcome to NextGen. Good morning. >> Hi everyone. Hi Neil, how are you doing today? >> I'm doing great. Thank you so much for joining. Really excited um you know, for you to share your story and share a lot of insights uh with the audience here. So, before we begin, I'll just you know, give a quick introduction about you, Rodrigo, so that the, you know, audience is able to paint a picture about uh you know, your experience and what you're going to share with us. So, everyone uh Rodrigo you know, has background in industrial engineering with more than a decade of experience leading innovation and digital transformation in mining. And he specially

**[2:09]** like specializes in systems optimization, mining 4.0, and industrial IoT. At SQM, he has helped turn autonomous drone operations into a very practical for monitoring irrigation across nearly 300 square kilometers of heaps. Connecting aerial data, AI models, and field teams to improve how issues are detected, prioritized, and resolved. Once again, Rodrigo, great to have you with us. Welcome to NextGen. And you know, really looking forward to for you to share your story today. >> Sure. >> Just a small note for the audience. So, the session is being recorded. I know that you will have a lot of questions. I sure will. So, please feel free to, you know, put them in the Q&A session. By the end of you know, the session, we'll take about 10 minutes and we'll make sure that we answer, you know, the questions that you've put in. If you find the session really valuable,

**[3:11]** and I'm sure you will, and if you feel that there's a chance somebody in your network will do, take a screenshot of the session, you know, post it on LinkedIn, add the hashtag NextGen. So, sit back, enjoy, and get ready to learn. Rodrigo, with that, I hand it over to you. >> Thanks, Neil. All righty. So, I'm going to share my screen. Please give me a second. Appreciate some patience here. >> [laughter] >> So, there we go. Neil, if you can tell me if you're seeing my screen, please. >> Yes, we can see it. >> Amazing. All right. So, hello everyone. Thank you to FlytBase for the invitation. Today I would like you to I would like to share how we use drone monitoring to improve our irrigation management at SQM iodine operation in the northern part of Chile. So, uh first of all, like uh Neil mentioned, I want to give you a little idea where we're located. Um we're located in Chile, to be more specific, in the north part of Chile, uh called

**[4:15]** the Atacama and I'm sorry, the Tarapacá region, uh the Atacama Desert. Our operation uh has a span of around 300 km². Uh like Neil mentioned, we operate a mine that uh uh extracts iodine from caliche. And we uh use uh leaching heaps to uh to extract this uh material. So, I'm going to give you a little idea of our process and how we use drones to improve our process. So, real quick, um I'm going to give you a little intro of our operation. So, um as you can see in this image, um after we mine, we place the caliche, which is the material, uh into heaps. So, what we do is this heaps we uh install this irrigation uh system with sprinkles. Uh then we start irrigating the heap. Uh percolation starts, then gravity

**[5:18]** brings the liquid down, and then we get our let's say uh a fluid called brine. That brine is very rich in iodine bearing salt, and that is sent to the plant, and then we produce iodine. So, that's a very brief explanation of our process, and now I'll show you where the drones help us to get uh the most efficient um uh to to improve our system and be more efficient. I'm sorry. So, here I'm going to show you a couple of pictures of we took before our operation started using drones. So, as you can see here we got dry areas and we got flooded areas. Our teams traditionally will go and check these areas by going on pickup trucks on this huge area and we'll try to pinpoint where we had this problems. That used to take a lot of time. You can imagine the resources that we had

**[6:21]** to spend to go and look for these problems and then when you find these problems, you then have to drive back because where we are located there's no 3G, 4G, or phone signal. You go back and put a work order and then go and try to fix this. So, one of the issues that we had we were very slow fixing this type of problems and this type of problems what will produce that will create problems on the irrigation and our yield will not be the most efficient. So, then what happened? DJI announced the docks, DJI docks. The first we tested DJI dock one, then we installed DJI DJI dock two, and now we're operating with DJI docks three. So, this was amazing. Okay, we found a technology that can help us, but now we need the connection

**[7:25]** to uh control all this fleet. So, then that's what when we found FlytBase. FlytBase help us to control all the drones that we had in our our field. Then it also help us to capture the images, create missions, and then send these images to our servers and help us get information from such images. So, >> [clears throat] >> this connection was just what we needed to put this puzzle together. So, I'm going to show you how we started doing this integration with FlytBase. So, we got the GI Docs on our field. Then, we needed a connection to transfer those images to our servers. Then, the servers get the image I mean in our servers we get images sent to Data

**[8:27]** Bricks, where we process with computer vision models. And then, we get a result to send information to our operators at the field, so they can go and prioritize inspections and repairs. So, thanks to FlytBase, we were able to connect the technology with the field. So, we started getting this type of images, where we were able to get okay, this is the thermal image and this is the you can say RGB image. And we with our computer vision model, we start to pinpoint where to go. That made our process way more efficient, and our teams only went to the field where we needed to be. Not to go there and search for the problems. So, I'm going to explain what we do right now, thanks to FlytBase. We get a review, a

**[9:27]** report to our operators. They go where they pinpoint where the damage was flagged. They go and check because it could be a a very easy problem or it could be a way more complicated problem, but still now we know where to go. We verify and then we repair. So, what did FlytBase help us improve? So, just to give you an idea. So, just in 30 heaps when we started this test with FlytBase, um it would take us around 5 days to go heap by heap and see if there was an issue. 5 days became 2 hours. All right. So, that doesn't mean we fix all the problems in 2 hours. That means now we know where we have problems in 2 hours. It used to be it would take us days, sometimes even a week if we had um

**[10:29]** a big issue and we couldn't send more people into the field. So, we saw a very very very strong improvement and I'll show you here with some of our yields. So, just to give you an idea, all the lines you see right here, the ones on the top are heaps that we started to attack as soon as we get issues. The only one that we didn't use the software is the one you see at the bottom, the blue line. So, this meant that okay, having a quick action thanks to FlytBase platform, our yields improved drastically. So, thanks to this, to our first 30 heaps that we were monitoring, management said, "Okay, we need this to the entire operation." So, nowadays, we got two drones operating, uh two more on the uh on the plans, and

**[11:30]** in that area and in the other mines that we have on the south part of of the region, we're going to at least install two more. So, nowadays, this is a tool that the operators really really really, uh, you can say, get mad when sometimes it's not something's not working. Let's say we have weather conditions that we cannot go out, they will call us right away, "Hey, we need the tool." So, that means that the operators really really like the tool and they cannot and they don't like to operate without this tool. So, uh, on the numbers, we got good results. And on the field with the operators, we got also pretty good result because now they really really want the tool to be working 24/7. So, that's a very very good, uh, you can say, something for us. Um, I've been flying those in in this slide, so I'm just going to I'm I'm going to I'm going to, uh, be a little bit slower so you can get what's

**[12:35]** coming in the future. So, all of this implementation, all of this, uh, system, uh, opened the door for new ideas, new challenges, and even the, uh, the operators at the field start to get in ideas, "Okay, we can use the drone for this, for this, for other stuff that, uh, happens on the field." For instance, I'm going to show you a little, uh, a project we're working right now. This irrigation line sometimes, uh, rupture because, uh, bad, uh, uh, bad, uh, let's say, uh, water conditions or fluid conditions or too much pressure and you need to go and fix this because a ruptured line will create, uh, flooded area that the drone will go and identify. So, at the beginning we were fixing problems, now we're preventing problems. So, and that happened in a span of what, 8

**[13:37]** months, which is for us very very um uh interesting. So, in the future we want to prevent the problems and send the people right away. So, uh we also have the idea to connect our flow meters, our pressure sensors to the platform. So, when a sensor sends data and that data it's not uh a reading that we want to see, we will send the drone and check uh if there's something happening around the area that we need to go and fix. So, we're very satisfied with results. We started from fixing to preventing and hopefully in the future we will uh be able to uh plant all the mine and have the drone as part of uh uh the process

**[14:38]** that will give us best uh yields on the field. So, basically for us the drone was the tool that we were always looking for. And without FlytBase and the ease of integration with our tools, we won't be as successful as we are right now. So, uh thanks to them. And to be to finish, like I was mentioning, um we want to keep working with FlytBase and keep listening to our team members on the field to still give us uh the best ideas and how can we always improve this um technology. Um the FlytBase team already knows that we're also looking forward to integrate this with security. Let's say someone wants to go and break into our operation, we also are looking to integrate this in our security uh division. So, yeah. So far in the mine section,

**[15:38]** we're very happy and we're expanding more and more and also um uh probably also in the mining or I'm sorry, the whole truck operation, we also want to be probably using some drones to make sure that the the mining operation is also working properly. So, yes. I don't know if um you have some questions, Neil, that I can answer. But basically, that's our our process. >> Thank you, Rodrigo, first of all, you know, for [clears throat] sharing your story. Uh you know, uh Uh I like to summarize it by the end of the session, but the chat section is actually fired up. So, I I'm going to throw a couple of questions at your end so that we can take them up. Uh so, when you started the session, you [clears throat] said that uh you use the DJI Dock 1, uh 2, and 3 currently. So, out

**[16:40]** of all the docks that you used, uh which one like obviously, the Dock 3 is being used right now, but uh which dock was performing good or better for your use case and why? >> Oh, a good question. To be honest, two and three were the best, but between those two, the only thing I would say we like more is the battery uh life of the DJI Dock 3. But to be honest, DJI Dock 2 had very, very good results. I mean, we used it for what? Um 6 to 7 months and no issues. Uh you got to add the the uh testing phase, which was three other three plus months. So, you can say almost a year of nonstop use. So, and the image quality was really good. So, to be honest, 3 I really like it, but the difference between 3 and 3,

**[17:42]** the only difference I would say is battery life. But, yeah, DJI is doing great uh work, I can say uh on that uh division of drones. Are you there, Neil? >> Yes, sorry. I had a uh network issue. I'm I'm really sorry. >> No issues. >> Were you able to answer the first question? >> Yes. Yes, I was able to. >> Perfect. Perfect. So, um I think one use case, and this is somebody who's asking like they're in the same industry, and let me just pull it up. All right. So, the question is that we operate in a similar environment, uh again, a large open pit uh you know, extraction operation in a remote area. How many docks did you need to cover the operation, and based on your expertise, uh what more areas of the mine could you potentially cover or, let's say, in the

**[18:46]** next couple of years? >> So, to cover our entire area, we use four four docks, DJI Dock 3, that will be able to cover our area. Um And the second part of the question was how many dollars are So, you're saying what more areas of the operation are possible to tackle with docks? What? So, I I'm getting that the question is what other areas we can tackle? Um I don't know if I'm going to answer the the answer correctly, but we're also thinking about to use it on the security and also in the mining extraction process. I don't know if that's the question, but yeah, we're using four drones and we're looking to expand it to the mine and security uh missions. >> All right. Uh there's another question. Let me just put it on the screen. How often are you flying drones? Uh is there a daily survey, a weekly check, or again

**[19:51]** like uh How how do the mission start? Like is there a trigger? Um you know, do you have a round-the-clock schedule for it? >> So, we got uh you can say three types of flights. We got the schedule missions, which they start at 9:00 in the morning, and then they stop at 1:00 in the afternoon. Then, we also have missions that are triggered. We're testing triggered missions right now. And also, we got missions that are uh manually operated in scenarios that let's say someone at the mine needs a an image or a video for a certain process. They will go and operate the the drone accordingly uh to their needs. >> All right. So, it's it's a mix of operations. It could be uh triggers, uh schedules, or you know, whenever the operator feels that they need to go out and fly. All right. So, let me pick a couple of other use cases or questions. Sorry. So,

**[20:58]** this this question is uh you know, from my side. So, initially you mentioned that you were creating work orders um you know, for the field technician teams. Uh how are you actually closing the loop? Let's say, you know, you're flying out on a schedule and the drone goes, it finds some problem, it captures the data, >> [clears throat] >> the data goes through your AI pipeline. Is that something that allows you to create the work orders for your field technicians on the maintenance team? >> Yes, so the cycle is the following like you were mentioning. We have the schedule flights every morning. Then the images will be uploaded to our AI model, computer vision model. Then the next morning the operators will have that report on their uh on their computers. Actually, they have like a little they have like a little um classroom, you can say,

**[21:58]** and they have screens of all the of the dashboards available for the operation and they will have a screen special specifically for this process and the software will be giving them like a top 10 and heaps that they need to go and check on the day. That's basically the the cycle and then when they fix the the the heaps that need to be fixed, they will create a work order and then that will be closing the entire, let's say, cycle of the of the issue. >> All right and all right, perfect. Understood. So, since we're already talking about, let's say, you know, field technicians, they can go out in the field to, let's say, fix certain issues that you found out. If I were a field technician and I've been using drones, let's say, at SQM, you know, how has my life changed is the first question and the second

**[22:59]** the nature of my work in the future as well. >> So, okay, the first question you in the second one you cut a little bit, but the first one how my work has changed. So, the first question I can say now you can go and uh do your work where the issue is happening, where the problem needs to be fixed instead of losing time going and finding the problem. So, uh your day will be more productive and you will only have to uh waste time driving because that's one of the biggest issues, driving. Cuz our area is so big, you only be losing time just driving uh to the problems that you're looking at I'm not looking, I'm sorry. The problems that we were detected.

**[23:59]** We detected. I'm sorry. And the second question, Neil, what was it? >> Right. So, sorry Rodrigo. Again, I had a >> What was the second question, Neil? >> I dropped off. So, I am My question was let's say if I'm a field technician, like how does my day-to-day change uh if drones are already being used as an infrastructure at the SQM mine? Like how does my life change in a way? >> Uh I I I answered the first part which just to just so you can hear it again is now you will attack the issue right away. You won't have to go and lose time looking for it. And then you asked a second part. I don't know if I answered that. >> Yeah, I think I think that covers you know most of it as well. >> Great. >> Uh we have a couple of other questions about other use cases. I think we've already covered that. Uh right. All right. So, I think

**[25:02]** covered most of the things, Rodrigo. I'd like to, you know, maybe summarize the session before we finally close, you know, for today. So, I think like the entire storyline that you told us like from the actual problem statement or the operational problem that you had to, you know, actually finding out an automated, repeated, and scalable solution so that you get are able to find the issues and again have a more proactive approach rather than a reactive approach. That's that's really amazing and it's really good to see that you've been able to achieve those results as well. And yeah, I think the end goal is to automate that pipeline, maybe, you know, solve more operational problems. Thank you so much for joining, you know, the session today and sharing your story and, you know, planting that seed in people on how, you know, a large physical site such as mines and integrating with different existing systems can solve, you know, big operational

**[26:04]** problems. Before we close the session, any thoughts or takeaways that you'd like to share with the audience? >> Yes. Um Just to summarize that we're very lucky right now that we live in this era. That I can help uh different um companies or different um institutions that are uh looking to improve the way that our workers spend their time. And it used to be that we lose a lot of time looking for the the opportunities to improve. Now, thanks to technology like this, we go right away and fix the issues that are happening uh in a matter that is the fastest. So, uh, if you want to contact me and get information of our challenge and how how we're uh, improving our systems,

**[27:05]** more than welcome to hit me with an email or through LinkedIn. So, yeah. Basically, that's what I want to send as a final message. >> All right. Thank you, Rodrigo. And I think the perfect, you know, you know, closing thought uh, for the session like uh, you know, a lot of people have a lot of friction to new technology, but the idea is that, you know, if you use it in the right way, you augment humans to make them, you know, better at their work, make them more efficient, and at the same time align them with the entire business goals of the company. So, you know, I'm glad that you shared that and well, you know, uh, that's probably one of the takeaways that I'll go home to and hopefully the audience as well. >> Yes, yes. >> we're almost up on time. So, thank you so much, Rodrigo, again for coming in, sharing your insights and your experience. Really valuable, uh, of course, for the audience. And thank you to the audience as well for, you know, staying tuned and learning a lot about uh, the

**[28:07]** different operating problems that a mind could possibly have. Uh, we have the sessions going on in the session as well. So, I'd recommend, uh, you know, you going to the main stage and picking the next session that you'd like to attend. Uh, thank you so much once again for being a part of this session. Thanks, Rodrigo, and we'll speak soon. Bye-bye. >> Have a good day. Bye, everyone.