# Deploying Autonomous Drones for Mining in Australia

- Event: NestGen ’25
- Published: March 7, 2025
- Length: 35:35
- Watch: https://nestgen.org/library/watch/6R6bWOkld7g/

> 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.

## Transcript

**[0:20]** hello and welcome to another insightful session at nes2 I'm nupur your host for this session and today we're diving into a real world World case study that highlights the role of autonomous drones in mining operations mining is an industry where Precision efficiency and safety are Paramount traditionally surveying and monitoring mining sites require manual inspections heavy machinery and a lot of time all while exposing Personnel to had hazardous conditions but with autonomous drones and AI driven analytics we are seeing a complete trans transformation in how mining companies can collect analyze and utilize data for smarter decision making to ensure smooth and engaging session here are some quick reminders drop your questions in the Q&A box anytime we'll address them after the presentation engage

**[1:22]** with us share your thoughts insights and key takeaways in the chat join the conversation on LinkedIn and Twitter using # nes2 let's keep the discussions going so it is my absolute pleasure to welcome Ravi sahu CEO and founder of stos and Ivan mckern the general manager at Rocket DNA unfortunately Cameron couldn't be with here however Ravi is a technology management executive with over a decade of experience in transforming technology business models to be responsive changing market and customer needs having worked globally in various leadership roles and have execute having executed large siiz multi-million dollar it engagements for Fortune 500 clients we also have Ian Ian is a mining engineer with a background in underground gold and base Metals mining he co-founded arvista in 2012 a business

**[2:25]** providing aial serving services to the mining industry since the acis of the business in 2021 by rocket DNA Ian has continued to manage rocket DNA's Western Australian operations please drop your posts and do let us know what do you think about the session on LinkedIn X Instagram whichever social channels you want to use a huge thank you to our sponsors who make nen 25 possible your support has helped us make NJ a success with this let's begin and I will hand over to Ravi and Ian to take it forward thank you nippur uh it's my pleasure to be here and uh sharing our journey about deploying autonomous drones uh uh in mining sites in Australia

**[3:26]** here so uh what we are seeing the industry shift here how the data is being collected utilized and analyzed while Doc is playing a role in making as a autonomy as a reality what we are seeing is the big uplift in the AI applications because you collect more data you need to automate the data processing need to automate the analytics as well and uh we're seeing a perfect balance of like how the AI is being utiliz in these mining operations so with that I would like to begin our uh presentations along with my uh partner here you know Ian who will also navigate this uh and explain this journey of the transformation in one of the mining sites here so bit more about isos what we do as a company uh we provide mining Solutions

**[4:26]** uh all over the world uh where our Solutions are deployed in over 40 countries in roughly 3,000 sites a lot of the applications that we focus is on mind to mail transforming survey data into meaningful operational uh analytics and bringing intelligence so starting from surveying to drill and blast and then geotechnical and then all the way to the downstream Mill intelligence but drone uh data as a key layer for that so as you can see starting from making the survey data go long uh further uh how to utilize that data to generate geological analysis then further extending it to the drilling and blast design planning and optimization using that digital Twi to understand what's the Hall Road uh conditions are and further extending to what's my blast intelligence is leading all the way to understand the Environmental

**[5:30]** conditions of the site using the vibration uh predictions as well but geospatial data is the key source for that so as you can see the dock as a uh bringing the autonomy we are seeing lot and lot more use cases and lot more data collection the enablement can only happen through the autonomy here so what changes here uh changes mostly is you have more data now so where we've uh bring in the AI as a layer is you have the data uh but what you used to do as more like catlike Solutions after importing and exporting into other Solutions now you can do more just by turning on this AI task or what we call AI agents as well to get more answers faster I'll hand it over to uh Ivan here to talk more about uh how they are running

**[6:33]** their operations when they're deploying the drones autonomous drones here thanks Robbie um welcome everybody so for those that aren't familiar with rocket DNA we are a global provider of drone Data Solutions uh we're an Australian based company uh with officers also in South Africa and Ghana uh with the the we've been involved in providing uh services to the mining industry here in Australia and Africa for about the last 12 years um so we've got a really good understanding of um of the industry and the data uh that we need to capture and what we need to be extracting from that data next slide please Robbie how uh the solutions that we um provide include conventional aial surveying services but the key

**[7:35]** focus of the business these days is the automated drone in the Box uh solution uh we've been doing that for about um a year and a half now and um so we're what we're doing is um building the hardware system and and conducting the remote operations um and then feeding that D with we're integrating with um with companies like stros to automate the uh processing and Reporting uh from that data next we the in terms of the hardware that we're using we're using the DJI products the DJI doc uh Hardware um as I said we've been doing um we've been operating these systems for about a year and a half or so now uh we've done uh many many flights over 10,000 flights now uh in mining operations uh in Australia and

**[8:40]** Africa um during that time we've operated in extreme heat conditions flash floods hail storms um and these these systems really stand up well to those conditions so we we're operating very consistently very very reliably capturing that data uh you know multiple times every day and um this integrated Hardware system that we've uh developed allows for the dock to be set up very quickly and easily on the Mind site next just um some more image imagery there sort of showing typical uh mind site uh setups um but again as you can see very quick easy setup next the operation of these systems is conducted by our Pilots um in their remote

**[9:45]** operating centers um they're they're doing the mission planning uh overseeing the flight uh operations so the majority of the missions that we fly are are automated so it's a pre-planned uh survey mission for example and then the pilot is um just monitoring that from a from a safety point of view at all times and if needed can take over manual control the key reasons in the in the Australian mining industry why people want to automate is um the one of probably the big one is the The Limited uh onsite resources so um it it's it's hard to get skilled people on mind sites in Australia which are typically in very remote areas and uh there's a very high labor cost associated with um with those people as well so um being able

**[10:46]** to do more work remotely uh has a a direct cost benefit but it also enables um very consistent capture of data so on a lot of mine sites in Australia it's the it's the mine surveyors that are out there flying drones capturing data um but then if uh a geotechnical engineer or an environmental uh person on site needs data then it can be difficult for them to get that data consistently uh when when when they need it so by um doing the remote automated capture um huge benefit in the consistency of that um capture and Reporting the Bas basic workflow um involves Mission the the mission planning which we which our remote operations team um is doing there uh many of

**[11:47]** these missions are um are pre-planned might be for example stockpiles which is the same Mission flowing every day or um you know these other pre-planned missions we we also do ad hoc um oneoff Visions as well the pilot does his pre pre-flight checks and Radio Calls remotely and then is monitoring that flight in real time the Drone lands uh recharges uploads data to a Cloud Server which um often is is our own Cloud Server um and then from there we can direct it to uh you know multiple other platforms so um often for us that data is automatically flowing into stos uh where the photogrametry and various reporting is happening in an automated manner thank youan um so I think you said the stage very nicely of like you know how

**[12:52]** to uh how to deploy the drones what are the operational uh capabilities are required to not only operate but how to do it right uh but I want to focus uh now more on once you have capture the data mve the data and process the data what are the newer use cases that can be developed you know from here what I call the AI power docks now uh and the use cases that we're going to see here in the next uh few minutes is uh how you can automate the Hall Road compliance uh what are the applications on making your high well more safer uh how you can monitor slopes more proactively uh and what it means for the regular pit analysis that usually gets uh carried on uh on a less frequent basis what if you can do it every couple days and then what it means for the Drilling and Blasting teams who need frequent data

**[13:55]** when they are blasting it before and after uh and then how does that impact looks like for mind development and the change management for your site here so starting with the hall road side the applications here and the impact is uh massive if you can monitor your Hall Roads and identify any physical geometrical issues that can prevent any increase where and teer on the tires uh if you are checking your grades regularly uh five uh automating this data collection and AI automatically identifying where their grade issues are safety bumps is a big really understanding where are the bottlenecks or where are the choke points are there and you can proactively go and identify so the data delivery to the reporting uh timeline becomes like very less and drives a lot of uh impact uh on a day-to-day basis to

**[14:58]** really understand uh this course Corrections in the Hall Roads so not only that you can improve your cycle times which means reduction in any traffic issues that scer uh reduction in the fuel and the tire use if your road uh conditions are proper and then uh of course on the safety side if you have uh maintained your road width properly there is you know less any incidents you know or about to occur there another application we see was on the geological side so why to monitor geological conditions uh on a more frequent basis is you can understand any uh conditions of where the dominant joints and false uh structures are what that uh data means is you can start to really good understanding of the whole structure analysis of your wall control which helps you in design engineering and planning side for your next uh uh bench control operations and how this relationship of

**[16:03]** different joints are developing uh which will lead to any uh safety issues uh and then further to that uh identifying any discontinuity issues uh how the materials the thickness and the characters are uh and then also the distribution of major lithologies here which means what different types of rock are there and all this is coming from uh the Drone data itself either in form of imagery or the other types of sensors and payloads that you have but uh the rapid enablement that happens in moving the data and AI automatically analyzing this uh uh reporting and the uh visualization allows the teams to mobilize and understand where these critical conditions are and distributing this data into different departments uh understanding of as I mentioned the lithology enables you to understand the different rock types are there so you the mines are chasing certain

**[17:08]** uh geological conditions uh they can clusterize those and they can plan and update their their blasting uh designs another application we see uh very frequently utilized in the mine operation is uh reconciliation as you continue to mine excavate rocks and extract your mind conditions and bench uh conditions changes what you have intended as a design and what's actually happening uh physically so the AI automatically capturing this information where those issues are so how my five face angle has changed where the catch bench width and issues are there so again reporting uh if you have the data understanding where those issues are and enabling those uh for the geotechnical engineers uh is a big help to reconcile the Mind plan uh one from design versus the

**[18:14]** act as build understanding where the slope stability issues are so safety risk factor analysis you can quickly see in the entire pit as you can see where my potential safety uh areas are where the potential risk are there red area showing the potential high risk for the slope failure so combining this data as well as the uh condition physical conditions it models and enables the team to like really understand so more monitoring leads to more understanding of the areas of the critical zones so how this uh enables from the side as well as the compliance side is we're continuously monitoring this information or this data is coming as in a form of the reality Capture Plus digital twin layer and the AI doing its job to automate this analysis can quickly monitor any wall failures uh or sterilization as you're

**[19:19]** continuously digging if you are over digging under digging those uh reconciliation reports are very useful but also it's like avoiding the rework if you are having a right data what's the conditions are on the field and if you continue to measure avoid any rework we've always thought in the other industry as you start to look for the manufacturing yit you have the entire s in the plans as well as in the construction industry what we call building information modeling if you're measuring continuously against those you can start to rework uh fast looking at the uh critical zones again from the over digging and under digging side what we call over break measurements uh and that's an issue that can lead to again ball failures uh not understanding kind of the geological conditions uh lead to again uh planning uh poor you know blast designs so having this information

**[20:25]** for the different teams enables lot more optimization for the mind stet uh which leads to kind of again maintaining the compliance but also improving the Mind designs continuously as well as minimizing any or losses that happens uh but more importantly your stability of P continues to strengthen as you're spotting this you know through the continuous flow of the data that's com in another application that we see where uh autonomous drones are uh very very kind of uh application driven is uh the frequency of the data that comes in so if you're blasting it regularly which is uh on most the Min is on the daily basis you blast you have a pre-blast and then the post blast so pre-blast you capture the drilling information Where to drill or understanding of where the qqc on the drill hose uh planning of the block self uh and then post

**[21:31]** blast the intelligence of what blasting measurements are there what how much rocks has come out but also what's that uh oversiz ratio there What's the total size distribution of rocks is and then uh measuring as uh what we call the back Break um of on high a lot of broken rocks are there uh that means there potential loss of O there so meing that before you start digging and planning for the next lost will lead to really good kind of understanding of continuous Improvement and then also the floor control uh usually Minds maintain uh uh proper floor grade that they have to drill to but if they are going uneven so you could quickly spot check as you are flying before and after uh which leads to a good understanding of where are the hot spots where these floors are uneven and this uneven floors can lead to lot of another damages to the equipment uh so your

**[22:37]** geospatial data which is from the Drone is have a direct correlation to the equipment performance now and then leading to kind of you know several benefits on the fragmentation optimization which means we can understand what's been produced in the mine the size that has a direct form whether the size that's coming in whether the crusher will accept that or not the shape of the Muk pile which is the blast pile whether our operations will dig that uh equipment will dig them properly or not and then the throw of the material which means whether this throw uh that is from the last perspective whether this is proper or not and then continuous tracking of the excavation leads to really good understanding of the non-compliant areas measuring whether were we leading to our production properly or not from high grade low grade to waste uh perspective and then quantifying your continuous changes on mind

**[23:39]** planning and pit development whether we are on target or not can be uh measured very uh easily here so there are various uh solu uh applications we saw from the AI perspective but we have seen uh a big uplift in the mins you know from their equipment performance cost reduction uh but as you can see 5 to 10% site throughput increase 50% reduction in oversized delay year uh but then overall we see 30% like reduction in deviation from planned what they are doing the planning as well as like what's been actually achieved as built here I will hand it over to uh Ivan here to talk about the actual use cases in one of the mines where the drones are deployed here thanks Robie so get one example of of a a typical mining uh deployment here in Western Australia um this particular site uh is located in a very remote location in

**[24:43]** the northern part of Western Australia uh there's no towns of any significant size in the region it's um one you know well over a th000 kilometers uh from Perth um so there's a very high cost um for mind sites like this to maintain an onsite fly in fly out Workforce um the this particular customer was Keen to uh maximize the amount of technical work that that could be done remotely from Perth which is where um where all the to the workforce of b um and and um they they were doing some basic aerial surveying uh on site just conventional drone uh work it all really came down to one or two surveyors that were trying to

**[25:44]** capture uh data for um a range of different people and applications that was um uh that was becoming difficult so what we did was deployed our survey bot which is our integrated system that uses the DJI doc 2 and we operate that remotely for them 7 days a week um right through daylight hours each day um we we've then integrated uh that system with stos for the automated photogrametry processing and and Reporting so from the from the uh from our survey B the dock the data is uploaded into our site Tube app um that's like a um a platform where the customer can view and download any of the data that is captured um by the system um and then in the case of any survey type missions that are

**[26:48]** conducted they flow directly through into stos um so in this um on this site the key key uses for the system is for um daily stockpile uh measurement um pre and post blast surveys um um and quite a bit of geotechnical uh Sur surveys and Analysis as as well so um in addition so for the I guess for the stock Pils and for the pre- and post PL surveys we're doing conventional uh Nadia sort of mapping missions for the geotechnical work we're doing very low oblique image capture of the um of the pit walls up very close high resolution um to generate the 3D models like the one that you can see there of of a slip in the wall um so what this system has allowed them to um minimize the number of people on

**[27:52]** site that um uh they were looking to increase the number of uh drill BL Engineers on site and um once this was implemented that wasn't um wasn't necessary um the a lot of their geotechnical work is done by a consultant based in Perth so that person is able to request emissions through a remote Operating Center and can get that geot technical information that he needs in a timely manner in terms of the in terms of the future um I guess one of one of the challenges has been the the uh the regulations involved in operating remotely piloted systems that's changing it's it's improving um the capability and reliability the hardware has come a long way in the last couple of years so you know our experience with the DJI do has been extremely positive um they're performing um

**[28:54]** very very well in these harsh mining conditions um and then the low cost of both the hardware and the remote operations is allowing this type of technology to be deployed at scale so on on you know it's quite feasible for a msite to have multiple of these units operating um the advancements in Ai and sensor technology um sort of leads to that reduction in in processing time and Analysis time and cost um and I guess looking further ahead there is the potential for fully autonomous systems uh it'll take a little while for uh the regulatory bodies to catch up there but um you know it's it's sort of Hing heading that way awesome thank you one uh for uh sharing your insights on the uh uh what's the actual kind of you know customer benefits were and how they have like seen uh and with that said uh we will end this presentation here and open up for any

**[29:58]** questions right so thank you so much Ravi and Ian that was indeed very insightful and I mean just seeing how Enterprises are benefiting and seeing the value of using these uh automated AI powered Solutions is is it's great to see and you know um I'm pretty sure the industry is just moving forward in the right direction uh so we have a couple of questions from the audience that we would like to take okay so the first question is uh what are the biggest barriers to achieving full Automation in mining drone operations and how is the industry addressing them I I guess um in terms of full automation I'm guessing this is sort of heading uh talking more about a fully autonomous system potentially without the pilots being in the loop um I guess as far

**[31:01]** as that is concerned it's um probably the big thing there is the regulations around uh how the these remote systems are operated so it'll be a while before um the regulatory bodies would allow us to take the pilot out of the loop um the technology probably still needs a bit of uh Improvement to get to that point but it's certainly uh C certainly heading that that why right thank youan um the next question we have is um how do you ensure data security and integrity while collecting and transmitting aial data from remote mining locations yeah I'll uh I'll take that and maybe if you have more kind of experience feel free to you know uh share that uh but I think in terms of the data security um most of the uh uh most of the data security kind of you know a lot of the

**[32:04]** application that have been deployed as a part of the doc are fairly secure so you are uh capturing the data uh storing it at the edge uh so that it's a lot more local you know servers are there which are essentially in client's premise uh uh you can use either client's Network so it's essentially going through the client's Network and then moving the data to the uh the actual centralized Cloud systems so as long as those are like fully the data transfer layer is encrypted uh as well as like it's authenticated properly and it transfer wi very secure layer the security uh measures you know uh can be Tak uh or concerns can be taken away there uh and at the same time the Integrity is one of the data Integrity which means how good the data the quality uh I believe that's kind of you know the initial Sops can be developed which um you know rocket DNA helps a lot of the I me developing like what type of you know

**[33:10]** data is needed and how frequently you need it either from a geotechnical perspective or just the blast videos these are two different things geotechnical perspective like you know you need to have a good quality resolution so those Sops are very much needed customer sure that when the data comes it's ready to utilize by AI Solutions more properly right uh sure AI uh thank you so much so we will be taking the last question as we're running out of time are you planning to also use synthetic data for training the AI models we have a question from NE who's uh asks if uh we're planning to use synthetic data for training the AI models yes so we use in some cases synthetic data uh for making sure that uh we are as much as we train like what exactly uh is there to identify and detect we also need to train the false positives uh so that it doesn't the model

**[34:14]** perform well in the false positive conditions as well so for that we create synthetic data to augment that uh and uh there is not in in the mining data uh conditions produced using synthetic data is little challenging because uh you need a lot of domain knowledge uh which means if you're a GE geologist or geotechnical Engineers what uh good data means and what bad data means it requires an element kind of understanding of what that domain is so uh there is yes we do augment the synthetic data uh in certain cases to uh Advance over uh the model training perfect okay so thank you so much everyone who uh joined in the session we will be sharing the recording pretty soon with you guys thank you Ravi and Ian for spending some time and helping our audience gain more insights uh please check out the agenda for more other sessions and uh we look forward to having you throughout the whole event

**[35:19]** thank you so much thank you and thank you everyone bye bye