# Securing a Classified Defense Site with Autonomous Drones and Robotics

- Event: NestGen ’26
- Published: October 2, 2026
- Length: 26:49
- Watch: https://nestgen.org/library/watch/j7KfY7PaJjk/

> 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

In this NestGen '26 session, MSS Security shares how they brought autonomous docked drones into a defense ground facility in Australia, one of the most restricted operating environments for any drone program.

Defense sites operate under strict rules around what technology is allowed on the ground, who can access footage, and what airspace the drone can enter. The session covers how MSS worked through those requirements and what the approval process looked like from start to operational flights.

Beyond the approvals, the session gets into the practical side: how the drone fits into a security operation that already has guards, cameras, and sensors in place, and what changes in the control room when autonomous aerial response becomes part of the workflow.

## Transcript

**[0:01]** So, Gregory specifically is talking about uh securing a classified defense site with autonomous drones and robotics. Uh to begin the session, uh let's start with like uh where people are uh joining us from today. Your city, your country, uh anything u is okay just to help us understand where everybody's from. Uh secondly, uh I hope that like I'm sure that like there's a lot of people who are uh tuning in specifically to understand like how a high value defense or crit critical infrastructure site actually works and like what are the top concerns that are actually being uh uh addressed by dog drones, right? So like go think of everything right from uh undetected perimeter breach uh to false alarms, too many false alarms, slow incident verification, poor visibility across such a huge site, um compliance, audit readiness and proving exactly what happened like a lot of times when an event happens, it's he said, she said. Uh right. So like proving exactly what happened

**[1:02]** with proof is also equally important. Um and that is exactly what today's uh session is about. this these kind of challenges when how do you harden the outer layer of a sensitive site detect and verify threats before the parimeter is breached and create a clear and auditable response chain from uh alert to close uh so we are looking at a real world security stack built around DIIB operations ground robotics and AI enabled monitoring as well as like 24 24/7 remote ops uh model uh this is especially relevant for people as I said who are joining in from like um uh other properties of like high value assets, classified information uh properties, export control technology or any kind of critical infrastructure responsibilities uh in these environments. Of course, like it's not enough to just have like fixed cameras and that is where Gregory comes in uh with his experience in the drone industry. He leads the aviation security and emerging technology at MSS Security. uh

**[2:05]** he's accountable for security operations across more than 20 airports nationally and uh he covering the entire operational performance, regulatory compliance and workforce capability in environments that run 24/7 and operate under constant scrutiny from regulators, government and airport operators. He also leads uh MSS's advanced technology portfolio across drones, AI, Motos and digital compliance. In today's session, Gregory is walking us through how MSS is approaching security for a defense contractor site using a layered model. Uh, and super excited to have you here, Greg. I think um, uh, we can dive right into it. I think we are like couple of minutes late to our session. So, I would recommend we dive right into it. What do you say? >> No, sounds good. Sounds good. And thanks for the intro. U, [clears throat] glad to be here. Do you want me just to fire away or do you want to point me in a direction? Sorry. >> Do you want me to uh just start or do you want to >> Yeah, maybe maybe like uh

**[3:07]** some context on like the site and like what how long have we been working with them? What is the deployment all about and then we can get into the solutions aspect of it. >> Yep. And uh feel free to tell me if there's any issues with the microphone. If you can't hear, it's coming through my webcam. So had some technical difficulties. >> Uh okay. So uh MSS security is a manpower solution um provider and what we're doing is we're also uh now looking at putting uh technology around to I guess enhance our the operational outcome. Um it's uh the nature of this is where I can't actually tell you the exact name of the site or the company that we're working in but it is a defense contractor working on defense land. So when it comes to all disk privacy um compliance everything like that it's a a higher level of scrutiny when it comes to operating in in these sectors but they also do have the demand it's just a much harder um I guess they can't shop from as many uh from as many places uh given given the restrictions they have on um the

**[4:08]** where the where the technology has been procured from um and uh and even parts within the technology itself. So we started by trying to supplement our manpower um with some AI that we were putting onto existing cameras that were working um different types of AI use cases that were happening for uh detection of people fire um and different different types of uh things like that. different sites especially defense given the current um well kind of a continuing um climate um any anywhere that Australian defense force is supporting either directly or indirectly it generally it does get some scrutiny from from members of the public and then those members of the public come and try to deface or damage property um which is defense or defense adjacent. So as a result they had to bolster their and really harden their external security while everything was really robust for their tight inner uh inner sanctuary of security that fence line perimeter and that outer layer of

**[5:09]** security. Um they wanted to go even higher. So for that it was manpower then we used the existing CCTV for AI capability and then we wanted to expand it out um much more. So then they um wanted to look at the drone capability that we already had. We were using flight base with the DJI drones and and that as we all know it's they're fantastic drones and the capability they have and um the functionality just keeps getting better and better. >> Okay. >> However, we couldn't actually use it. We couldn't use that because it's a Chinese um made brand. So we had to look at a um a USmade or blue chip type um drone. So, if you wanted to flick on to the to the next one. >> So, we went out into the market. We're looking at what was available, what was going to be, again, we're a business, so we're looking at what's going to be a cost-effective option for our clients as well. We want to be get best bang for your buck um with really having that uh that that secure product in the end that can deliver. And for us, um Skyio was was really there was no question. Skyo was the one

**[6:14]** for us. They had their their drone in a box. They worked with us really well to be able to initially put a trial in for the client. Um, and it was really quite seamless getting it there. Their technicians are really supportive everything um the whole way through. We were able to uh set up relatively quickly with ease and then we were doing perimeter security to start off with. So we were within a close proximity to an airport when we were doing this. So it did make it difficult with compliance wise to be able to to fly. So it didn't come under our broad area approval. We had to actually start off doing an EV loss. So we flew it from our remote operating center in Melbourne and then on site we had a spotter and we were able to actually start flying that. That's how we were able to commence immediately. We were doing our patrols and then we were able to actually respond to alarms um sensor lights, anything like that that was happening. The guards were able to contact us and we were able to immediately uh deploy the drone or redirect it. it was on a fly fly charge, you know, as you do with your BB loss operations. Um,

**[7:18]** really great functionality with everything they've got. It is they offer quite a lot like like DJI. It's just um you you you pay a significant amount for it, but we look we had a really great experience um with them. Uh and it is something that now is going to sort of I guess push out and it's we're going to be doing a significant uh bit more with it. So with this there was a whole bunch of layers we had to go through with cyber and making sure that everything for the client was um up to scratch and that the fence were actually happy with what was going to be happening on their side. So we were able to tick all the boxes there and then we went out and we did a few months patrol which then eventuated into them wanting to put this in fulltime into their site. Um and then again now we're looking at putting it across more sites which has been really good. Um but what we found was that there was still a gap While you can have people on the ground, unless you have a large number of people, which is costly, we're always trying to find that where can technology not replace but supplement the people to really provide that more

**[8:20]** effective but also safer outcome for where we can. So then then we started the discussion about uh about dogs and it was about real dogs um real real attack dogs and sniffer dogs and things like that and looking at what we could do with that which then I guess moved along to robotic dogs and uh and then that's where we sort of got on to setting up we already had an existing uh robotics just vertical that we were working with uh they were in the cost effective model so it was the Chinese uh dogs not the not the blue chip dogs um that that we were trying to work with um for our other non um non-effective clients that we could use um different types of uh technology for and still achieve the same outcome. Um but we were required to do a specific type of um robot for for the client. So I think if you go on to the the next one. >> Right. >> Um so again still going through uh a significant amount of Is that the

**[9:22]** the one before that? >> Oh should I go back? >> Oh yeah this one. Sorry. So in the air um Guardio could do quite a bit. It could see quite a lot but there was a bit of a gap. The drone did have to charge. So we were thinking about a model where we could deploy deploy robotics uh to be able to fill in that gap. And while there are cameras and there's quite quite a lot of coverage that they've got, it was really about having that that awareness and the flexibility to be able to go and investigate anything that we wanted to uh that we worked to find. Um our real concern was that members of the public would see this through the fence and like the drone and then they would think it's attractive and more people would come to the fence line and it wouldn't really have the desired effect. Um, so it's about uh really really positioning yourself on uh how you're going to be doing patrols with the drone and with um with the with the dot to be able to minimize the attraction that you're going to be attention you're going to be getting brought from external um people. So we had to do an assessment through that

**[10:22]** and work out how we worked at heights and like the way path in which we were going to be deploying this um robotic um robotic class. >> [clears throat] >> So same thing the drone would take up to unfortunately in Australia sometimes up to 12 months for an approval for a BB loss approval. Um so that is the reason that we moved into uh the robotic option so that if somebody had a need for a portable camera effectively a drone we could have a groundbased drone and we deploy it immediately and I believe FlytBase are potentially working at different integrations and things like that with different types of uh hardware which is really exciting. So, I'm really keen to see what what's uh what's what's happening in that space. Um because yeah, the the robot dogs that you get at the moment, they're pretty raw and you need to actually program quite a lot into it to be able to to get anything out of it. It's it's not not quite as easy as you would uh as you would think. So what we did, we've got our AI product, we've got our drones, we've got our robotics, and to actually connect them all in together. We're able to connect it through our one

**[11:23]** interface, which is secure, which is similar to, I guess, the product that you have with your with your internal AI, which is which is, you know, we do use when we're when we're using the FlytBase software. And we're able to get drone footage coming into our AI platform, uh, robotic footage coming into our AI platform, live CCTV footage coming into our platform, body warn cameras on guards coming into the platform, and everything is centralized. So nothing is missed, everything is recorded, and it's all compliant with um with with DIS, with our ISO, with essential and everything that you need to um abide by to be able to function in one of these restricted type areas. Thanks. So, so that's really I guess where we went. We went, we tested it. We actually did we we found that when we've done drone deployments in the past, if I [clears throat] digress a moment, you can have a drone up there for a trial, for an implementation, for whatever it is, but sometimes nothing goes wrong. Nothing goes wrong in the security sense or the safety sense. So, you don't have much to report on. It's

**[12:27]** really hard to justify the existence of of the product. Um it was quite the opposite with this and we actually had some sort we've actually had situations where we've been able to find things that have happened um not directly to our site but to our own neighboring sites and things like that which has been quite quite good. >> So we're able to actually engage with all of our all of our tools and and really see uh what the benefit is with having that limited exposure to our to our own side as well. So, it was a couple of good uh good situations that we were able to get some really great results from, which is able to justify um what we're doing and our investment and then we're to push out into more of the defense space and um you know, aiming to be able to um get larger and larger in that that that arena. We also work within um critical infrastructure as well. So, we do have a couple of clients that are coming on now with with some drones and and uh and AI and different types of um applications. So again, uh, all this type of equipment is what would go into there. Unfortunately, DJI can't be used at um, all of them, some some some can. Um,

**[13:29]** but until we were able to secure that back end, I think there's that there's risks with that. So we found that our product secure, again, it could be any product to be honest. It could be your milestone, your genetics, whatever it is that you actually have, which is your FlytBase. That's why we're here. It's actually the flight base can actually do all this. Everything could feed directly into Firebase. You could run it through AI. Everything's in one location. We have a 24/7 rock. We have a 24/7 monitoring room for the alarms and the cameras. And they are separate rooms. They're not the same. So, um, so they're able to actually fly. They're able to capture anything. Anything that the pilot doesn't see, the AI does pick up, sends the alerts through, and then we're able to actually deploy either people on site, law enforcement, or our patrols to actually be able to respond to it. So, we've had uh quite a bit of luck with uh with what's happening and um just keeping to hoping to just sort of continue uh doing what we're doing. It is u it's an awkward presentation I'm doing because I can't tell you the client. I can't

**[14:31]** tell you where it is um and tell you a little bit more about sort of how we came from where we were to where we are today. Um however, it's um it's working really well. Sky has been really supportive of everything and uh and yeah um if there's any questions [clears throat] I' I'd be happy to to go through those and uh sorry sometimes I skip through things a bit quickly you're on you're on mute can't you if >> sorry am I audible >> yes so thank you so much Greg I think that was quite insightful and as you mentioned like of course like while we can't share a lot of like the details of the site and the deployment etc. but I think just like it's it's actually quite inspiring for everybody in the room to just see that like the scale at which like a defense customer is adopting um this technology and especially like what you guys are doing with AI now. Uh, of course like the unified monitoring thing has been uh

**[15:33]** around for a while but uh what you're doing with AI and like I I read what you had mentioned that like um three out of four four alarms are just you know filtered out by AI. So I'm sure that that's creating a lot of value uh for the operators on ground. Uh so super inspiring and I I think we have a couple of questions from the audience as well. So if you don't mind um answering those I could just like display them on the screen for you. Yep. >> Okay. So, the first one over here is uh when the drone detects something on the perimeter, who gets the alert first? Is it MS Security's operation center, the defense facilities own security team or both of them? >> So, it actually can come to anybody the way the platform works. So, depending on who it's not actually defense, it's the defense contractor that manages the area. So they have access to all the alerts coming through on a mobile app on their phone and we have a 24/7 monitoring center which is consistently monitoring it. So they'll be

**[16:34]** the one to contact. However, um the client actually has access to the to the gooey and everything. So they could be they could be receiving alarms at the exact same time and and watching all the alerts and everything at the exact same time, logging into their cameras and watching the live footage >> um as it may be. And then they communicate directly with either the pilot or the person in charge of the robot or the monitoring room. uh about what they actually want to do. >> Understood. Uh so the next question uh we have here is defense clients tend to have very specific requirements around data sovereignty and who can access footage. How does MS handle that when running the system from their own ROC? >> Yes. So everything um everything does go through and meet the standard with u in conjunction with Scottio and how how that footage actually comes into be and then with our with our secure right platform and all of our footage um with our ISO accredititation we have a disc certification as well. Um, and

**[17:34]** there's a couple other things that the the more technical people should have been here to actually explain to you on how we actually have it, but there's a whole lot of boxes that actually need to be ticked with uh when it comes to the data sovereignty and data security. Uh, and also the time actually keep the data as well. So, um, that that was probably the the most >> painstaking long um part of the getting the drone there and flying it was the easy part. It was >> it was making that everything was u was secure and u and again what what we're looking at >> um it's the it's the less side of less side of security if you would. It's not um we're not going in there and looking at you know military operations that are happening which would be a higher level. It's it's on the edge of it. So it does come under a different level. However um we're compliant for all of it anyway. Absolutely. And I think we did touch upon this point like during the keynote as well of next gen that like flying the drone is like not the hard part anymore like the tech works but like figuring out like compliance for so

**[18:35]** many different types of sites and customers like working with their constraints uh like is is the uh problem that like most of drone pioneers are solving today especially like security companies like MSS. So, it's really interesting to see that like uh it's actually coming to life and it's not just like a proof of concept anymore, you know. >> Yeah. We've actually had to hire a team purely for um for the the back end. So, not not just the drones, but about data and and and the footage, the large amounts of data that capture everything. >> It's it's it's really getting out of hand just [laughter] how much everything has. I completely understand it. It is just it is it is becoming an enormous part. And um I guess from my my fear being that we we we're a provider that actually sells products to people and like like probably many of you the the more intense it becomes the more cost is associated with it. The more cost is going to get passed on to the client >> and we don't want to have things that are going to be cost prohibitive to to do these innovative ideas. So it's it's trying to

**[19:36]** find that that medium where >> you either get enough volume that covers the cost of doing it. Um, but what we do given our given our size as we just we can take a hit for for a period um and it doesn't really hurt us. So unfortunately the smaller companies can't can't do that. So I'm fortunate in that regard it's not as much of a a hit but uh but yeah it's definitely a costly uh process. >> Understood. That's quite understandable. So, so along the same lines, uh now we have another question that what level of vetting and approval did MSS need to go through before an autonomous drone was allowed to operate on a defense especially I think you mentioned that the site was also very close to an airport. So I think uh like that must have added a lot of challenges to the BV loss approvals. Uh so yes so the when it came initially so initially to get it off we did EV loss um and it wasn't a restricted area so it wasn't a restricted flight space which a lot of defense bases are so this was a um I said defense adjacent base uh you would call

**[20:39]** it where sort of the land's owned by them but it's ran by other people >> and um >> and so so yes it was getting approval through through our clients and what we needed to do and they went through the right processes to get approval to do whatever we we could do and at the same time um then we needed to appease you know CASSA and their services and everybody to make sure that um there was no issue there and there's definitely been some limitations with what type of approval we've been able to get for BB loss operations um the heights and and and and timings and different things like that there's restrictions on um when we can go it's >> it's not too bad but yeah and it's it's based on it's based on where you are um the number of flights that are coming over you and different things like that. >> Absolutely. Uh I'm going to pull up the next one over here. Uh defense facilities have some of the strictest uh security requirements of any site type. Uh what level of vetting and approval

**[21:45]** sorry unless they're mentioning so everybody has to have the right security clearance before they can go onto site. >> Yeah. um installers, everything like that has to tick all the boxes and that that's kind of just pa for what what we need to do when it comes to any any defense related anything. >> Uh yeah, I don't know if that's what the question was referring to that >> but yeah I think no it was along the similar lines what we touched upon like u that like uh to even fly on that ground like there are there was quite uh some challenging work in terms of security approvals to be done. Uh but we can I think we um can move on to the next question over here. How does the autonomous drone patrol integrate with the existing security infrastructure on the defense ground? Like they could have fence sensors and CCTV and access control systems. So what is I guess the question is like how are you integrating everything into one or are you uh in the first place? >> So we we're able to bring it into our our platform. We can bring in fence sensors on um like

**[22:45]** products like FFT, which is one of the one of the good ones that's out there at the moment. We can get those alerts that come in and then it sends the exact location of where the where the drone needs to go. So, >> um I believe you can put into a flight pace. You'd be able to put into flight pace quite easily as well and then you're able to deploy the drone. So, you get your signal of the location and you're able to um get that sent to the drone. So, that's quite uh quite easy there, >> right? But secure specifically is is more like a VMS. Correct. >> Yes. Yes. So it's a complete AI platform. But the the VMS where it can it's it's agnostic. We can plug in we can connect into um you sort of VMS software or we can connect directly into the IP of the camera and take the feed. So either either we can take anything that has access to the internet we can take into our um into our um interface and then we can run AI over the top uh quite quite seamlessly depending on what we actually what we actually want to do. Makes sense. Makes a lot of sense. Uh, all right. I think like that kind of answers

**[23:51]** all of the questions that were in the Q&A tab. If if you guys have any other questions, please feel free to like connect uh with Gregory. Uh, what I'm going to do is uh I'm just going to quickly send across um Gregory's LinkedIn as well. If anybody wants to connect with him after the session, it should be easy for you. Uh and of course like with uh the kind of experience that Gregory comes with in the drone industry, it's uh quite some insights that uh we can take from him. That's um so yeah, that's Gregory's um LinkedIn in the chat box. Uh feel free to connect with him. Of course, he's here for a couple of minutes if you have any more questions. >> Sorry, Greg, you want to say something? No, no. I was just going to say if you do have questions about specifics when it comes to the back end, I will I will ask the right people the right questions and I'll be able to give you the exact [clears throat] direct on what uh what they do. I I don't pretend to have all the answers and I I employ people that do.

**[24:53]** >> No, but more importantly, I think like the whole idea behind NextGen is to like build this community that's like a global group chat of people, right? And uh like you never know like I'm pretty sure like people from across the globe have tuned in. Uh so um it's it's always great to sort of like work with to get to know like-minded people. I think Timothy is texting here. Would love to hook up Gregory and share knowledge and ideas. We'll reach out on LinkedIn. So Tim is also um presenting today. Um later today he has a session. Um I I don't know if you guys already know each other. Sorry. Uh, >> do you guys already know each other and turn? >> So, I can't actually see on the screen. Um, >> ah, can you see the chat box? If you're >> is it public quick Q&A or chat chat >> chat? Yeah. So, um,

**[25:55]** uh, No, >> I think uh no worries. I think uh Tim will anyway connect with you on LinkedIn and we do have a session with him today uh related to how god direct is running an ROC and how they uh ran a public safety event dog drone and a public safety event Apple pop in the Netherlands. It was one of the largest events over there. So I'm sure like you can tune in to uh that one as well. It's going to be pretty interesting. Um anyway, thank you so much uh Greg. I think this was a really insightful session and uh um of course like we are so happy to welcome you to the Nesten stage. Uh this session will be published later on YouTube and um like anybody who is joining in can always come back to the session recording for the insights. Thank thanks Greg. See you around.