# How Skyports Built One of the World's Most Advanced Autonomous Port Deployments with MPA Singapore

- Event: NestGen ’26
- Published: October 2, 2026
- Length: 31:31
- Watch: https://nestgen.org/library/watch/2mvWP4pe9oM/

> 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

Ports and maritime operators need better ways to monitor large, high-traffic environments where security, operations, and response all depend on faster visibility. Autonomous drones are becoming part of how advanced port teams improve oversight without relying only on manual patrols or fragmented camera coverage.

In this NestGen session, Skyports and MPA Singapore will share what it takes to build one of the world’s most advanced autonomous port deployments, and what port and terminal teams should think through when evaluating similar models for maritime surveillance and operational visibility.

## Transcript

**[0:00]** Hello everyone and welcome to another insightful session at NestGen 2026. Well, this is personally one of my favorite sessions because I've been a part of this deployment and it's very unique and challenging in its own way. My name is Neil and I'm the head of customer success at Tritebase and I'll be your moderator for today. So ports are one of those environments where the value of an autonomous uh drone dock operation becomes pretty easy to understand and see. But deploying them at scale is a very different challenge. You're dealing with huge areas, constant movement, vessels, people, uh a lot of security requirements and operations that really can't afford blind spots. And that's what makes uh ports such an interesting physical AI environment as well. You are talking about autonomous systems putting them into a complex environment uh which is const constantly changing

**[1:02]** uh and it has to operate reliably and that's what we're going to explore in today's session autonomous sport deployments with MBA Singapore. Before we get started I'd love to know a little bit more about the audience. Please tell us in the chat section where all are you joining from. Do you work in the autonomous sport industry? If yes, are you starting your program? Are you in the initial st stages figuring out how you can scale everything from the ground up? I just have a couple of housekeeping rules as well. So the session is going to be recorded and we'll post it after the event. We have a specific dedicated time slot for questions and answers. But if you'd like to, you know, post them during the session, feel free to do so and we'll pick them up at the end of end of the session. And while we're doing the session, just hold one thought while um you know uh our speaker here gives you uh a walk through on how it's done. What does it actually

**[2:04]** take to move from a promising drone use case to an autonomous operation that can work reliably in a port environment? And we are going to hear directly from the experts. And this is where I'd love to introduce Jeff Giddens from the Sky Ports team. So Jeff is the head of technology at Sky Ports. Welcome Jeff to the session. Really excited to be hosting you today. >> Thank you very much. Great to be here. >> All right. So and yes, if you find the session valuable, please let other people know. take a screenshot, post on LinkedIn. And yes, without any further ado, Jeff, I'll hand it over to you so that you can share all the insights that that you have for the audience here. >> Great. Thank you very much. I'll share my screen. Should be able to see that. Um, great to be here. Thank you very much, Neil, for for

**[3:08]** inviting me to speak today. Um, I'm going to tell you more about how we worked with MPA to design, develop, deploy, uh, and operate, uh, a drone service for them on a daily basis. Um, we've been operational for over a year now. Uh, lots of lessons learned along the way, and, uh, I'm going to tell you in the next session how we did it, what the key, uh, things were that we picked up along the way, and, uh, how we went about it. Um first a little bit more on Skyports for those that uh that are not familiar with what we what we do. So we are a drone service provider. We are active in three main verticals. So we do inspection um both from a drone in a box monitoring capability for example like construction monitoring where we do a lot of like these massive construction sites where they're building for example sports stadiums. Uh we also do power line inspections both with DJI uh based drones as well as non non-Chinese uh drones as well for those uh uh customers that have a have a more sensitive nature. Um we provide surveillance

**[4:10]** so like for what we do for uh in Singapore we provide docked coastal surveillance. So we provide docks near coastal areas or like ports and we also do uh long range maritime missions as well like 8 hour plus missions uh for our customers too. And the third main vertical is is delivery. So we do uh medical logistics for example between hospitals uh in Europe. And then since this summer we also started offering offshore delivery. So from operational support vessel to the top of a wind turbine for example and we use a DJI fly car 100 for that. So those are the main three verticals. For MPA in particular um the focus was just surveillance. So we provide a complete surveillance service. So we're doing uh figure out what they need, how to deploy it, how to operate it and then turning that data into useful intelligence for them. Um a little bit more about how we went about it. So uh MPA came out with a tender uh I guess in spring of last year um in which they defined quite loosely like okay this

**[5:11]** is the kind of coverage we want the number of boxes that we want um this is the kind of APIs um that you we need access to and that made it very clear immediately for us that uh the the standard DJI solution was not uh what we wanted to go and offer and that's that's basically what brought us to FlytBase uh and and the partnership that allowed us to win this contract from from MBA. So we work with NBA first did multiple customer workshops to figure out okay where do they want us to operate from? What are the operating environments any kind of constraints constraints there? Any kind of regulatory approvals that we need in order to operate there. Um so we did a couple of customer workshops came up with all the requirements uh came up with a design solution and figured out okay these are the we did on-site assessments of the dock locations as well. Figured out that some of them didn't have any power available. So we had to come up with a a solar power based solution as well to provide uh coverage throughout the day. Um and then there were other requirements like for example the customer wants us to build a remote operation center within

**[6:13]** their offices. Normally we operate from our own remote operation center in the region. Um in Singapore we have one of our own in Singapore but in Europe for example we have one in the UK that we operate our uh our flights in Belgium from. We also have one in Madrid. Uh and we have now in Australia one as well as well as in the US. Um so but because this is a government customer, they want us to operate from their offices. So we basically took all our hardware and deployed it into one of our one of their their uh their office locations that allowed them to basically drop by, see our operations um ask any questions and uh also help with ad hoc deployments uh and man manage emergencies more efficiently as well. Um so back to back to the flow we did uh basically came up with the solution design got it approved from the regulatory perspective as well. Uh in parallel we set up the remote operation centers at their office location and then over time this is not something that was done on day one because like initially they just kind of wanted access to the APIs. They wanted to

**[7:15]** have the raw video feed and they want us to be able to give the customer and and the different people within the organization live links so they can follow along as well. Uh but over time we developed a custom dashboard for them as well where we moved away from just providing them with the raw data into actionable insight uh which is way more useful for them on a daily operational basis. Um Singapore is uh probably one of our most challenging operating environments in the world where in other locations it's often possible to operate up to 120 m above ground level. Singapore uh has a constraint that you can only operate up to 60 meters. Uh above that military airspace starts and you're also quite close to Changi airport. Um so that's one of the main reasons that you're you have to stay below 60 m. The big challenge there is that you have ships actually that are taller than 60 m especially the bigger cruise ships and the massive container ships. Um this is not just from a from a routing and or pathing um perspective a challenge but also it just can impact your command and control link. So we have a live situational awareness

**[8:16]** capability where we pull in the AIS information. We visualize that in uh FlytBase as well. Uh and that allows our pilots to keep aware of where the the big ships are, where they're going, where they're going to be, and also how that relatively impacts our own routing and the pathing of our of our drones. Um in the end, we opted for three different locations. We have two docks per site for continuous coverage. So when one is operational and flying the other one can be charging. Uh we have one of pajang uh that's the complete off-grid requirement. So we had a massive amount of solar panels in order to make sure that the drones stay powered throughout the day and we have battery setup as well. Then we call WCP that's a very constrained waterfront footprint with a bunch of obstacles nearby that make it more challenging. And then at MDE the marine uh drone estate where other drone operators fly Bville is from as well. It's a mains powered site. uh the challenge there is just deconlicting with oper other operators but uh there's like clear communications uh channels in place for that. So that's quite easy at the end

**[9:18]** of the day. So this is a remote operation center where we operate from um so we basically have redundancy across every critical aspect from uh networking to uh ground control stations everything is is still redundant to internet connectivity as well. um so that we basically uh are operational at all times. So even if the system completely would lose power uh we still have uh capability to at least bring the drones back safely uh and land them. There are two types of operations uh that we do for MPA. So there are the predefined routes where we're basically looking at um asset and compliance checks. So we basically go out, we check whether buoys are still in a in a decent state, whether all of the the markings are on them, whether the moorings are all okay. Um we check for uh flat sum and oil sheets and things like that. And then there's the ad hoc tasking support um which could be day and night depending on the the nature of the incident. Over the last year, we've supported anything from fires on ships to

**[10:23]** missing people um and support those operations with other government agencies as well. And the useful thing about FlytBase there is like because you can easily share links of live feeds, you can basically give it to people outside the organization that need the information as well, which has been really useful for us. So how we use flybas operating layer. So it's uh a remote command capability. So it's we use it for supervising uh the aircraft and the payloads. Um we have repeatable and predefined routes. Um at the moment we're also working towards one to many operations. So where initially there was a lot of manual navigation and a lot of manual flying like we just had a couple of basic waypoints as we were finding our way across the port that has grown into just like um um work where we have just predefined way points we have predefined actions at those way points and we're now uh operating most of our missions uh completely automatically uh where we doing the routine checks. Um we also have of course the live video telemetry the map overlays and then what's

**[11:25]** most useful is like the information that you can get out of the platform after the flight whether it's uh the images from the flight the metadata attach that to those images or where those were taken the flight logs trails um because the customer also wants to know okay what coverage of the port have you had in the last week for example so all that information can be plotted easily on a map and a dashboard and it's basically something that while initially of part of the project where we were doing manually And there was a lot of admin hours uh related to that. Now how it's just all automatically done after the flight and we provide as a as a dashboard overview to customer. Um here's an example of one of those uh ad hoc uh deployments that we did. So this was a a lithium battery fire on a vessel that had to be monitored for multiple days and at some point you couldn't see the flames anymore but with our thermal capabilities we could easily uh designate, okay, this is the containers that still have hotspots in them uh that we have to address. and we we shared that with the uh the Singapore uh civil defense force who was uh monitoring and and dealing with the fire.

**[12:25]** [snorts] Um so a little bit more about the type of uh missions uh that we do and the type of checks that we can do in port environments. So we can check assets whether it's buoys, moorings, navigation marks, we can check for compliance. So in Singapore that's uh ships too many ships mostly uh pleasure vessels that are uh too many of them are attached to the same mooring. Um ships that are for example uh obstructing the fairway like Singapore being one of the most busy ports in the world. Um the thing that any given point they have about thousand massive ships uh across the south coast and if any of those big ships are obstructing a fairway that can have a massive uh effect on on other ships moving around the port as well. Um and we also uh look for uh when there is restricted activities taking place whether make sure that people are being compliant with that and we look uh into environmental issues as well. So um we look for oil machines we we check for flat sam and uh we check for ships not following the right uh emission standards in the port uh as well and we basically gather evidence we provide it to

**[13:29]** MPA who then deals with that. So where initially it was all a video feed, it's definitely not that anymore. And we basically it's just the evidence gathering uh operation now where we combine the imagery, the location, time uh as as part of the evidence for MP to deal with. Um this is what the overall uh process looks like. So we start by capturing the information. We uh monitor the operation, contextualize it with additional information like AIS to find out where the ships are and how that might impact our our drone flights. We also use FlytBase to uh put in all of the the geo fences because uh port environments is uh very complex in that way to make sure that we don't go anywhere near where we're not supposed to be. Um we visualize all the information for the customer on the live map using the flight based APIs. Um and that also allows us to to track the KPIs like uptime for the customer as well. And that through that platform the customer can basically act. So they can track historical incidents, they can coordinate a response uh and they can they can track

**[14:29]** uh incidents as they unfold. So a little bit more about the platform that we developed. So this is all based on the APIs that are uh available from FlytBase. So in here um you can see the three locations at the center of the the yellow circles of where we operate in the port today. You can see historical tracks. This is something that the customer want to see and we can basically say whatever time frame you want to visualize those tracks can be can be made visual. So we can see last 30 days, last seven days but but also custom custom time periods. Um all the little stars are different types of um incidents that historical incidents that we were tracking and you can see on the bottom right what they all mean like oil machines or PPE incidents of people not wearing life fest for example fatam. Um, you can also see the uh uh the red circles with like the the yellow little markings in there that those are like ships that are in the port. We need to know where they are in order to avoid them or navigate to a certain ship as well. So, we can basically click on the ship, it will give us all the the AIS associated

**[15:31]** information. Um, if for example, we have to go and check out something that's happening on a certain ship that allows us to do so. And this uh platform also allows us to visualize the live telemetry of the drones as well. So the customer can look at this platform when we are actively flying in the port. They can see where the drone is and also there's a projection of what the camera is looking at at any given moment. Um so this is something that we developed for customer. we we start noticing that they were getting lost in all the spreadsheets that we're providing them. And this is a way more visual way of of of providing that information. And also this way you can see um different types of u how the data unfolds over time like different patterns of like where the certain oil streams tend to take place more where certain flot tends to gather. Uh it's much easier to deal with it that way uh in a in a visual kind of map. We also provide them with um automated dashboard. So they want to know how many flights we're doing, what our uptime is. Uh they want to see like an overview of like all of the the recent incidents and then be able to deal with them and and and

**[16:34]** sign them off as they as they go along. And this allows us to just provide all that information automatically. So we don't have to spend time uh looking at it ourselves. So you can see that over the last year excluding weather we had a 98% uptime which is pretty good uh even because some of the the drone docks are completely uh solar powered which uh introduces additional complexity and taking into account the weather because sometimes uh it does get go out of limits in Singapore. So we still managed to get 85% uptime which is pretty good for these docs. Um, we've also been experimenting with the the flybased workers AI capability, the kind of world model and there's so many different use cases to apply to in the in the port environment like automatically detecting oil machines, smoke, different types of objects. Um, and we're basically in experimenting phase with that. So we still have an operator uh in the loop that validates it and we've realized that we need to because our use cases are so so custom and unique uh we need to basically

**[17:35]** gather more evidence that we can then feed into the model so that the model learns gets better and that will allow us to get to those one to many operating operations over time where at the [clears throat] moment we spend a lot of time flying the drone still even though the the the flight is automatic the the operator looks at okay what is what does the drone need to look at what what do we need to uh sway our gimbal to and focus our camera on like going forward if we can automate that further with with AI tracking capabilities um that will enable us to do one to many operations when uh more airspace becomes available to us above the 60 m. So finally like what what do we get to and where are we going going to in the future? So we we have a proven service now above 85% even with with taking into account the weather. Uh we do routine taskings, emergency taskings um across three sites. Uh deconlicting is not not a a problem there. We're able to incorporate all the other sensor feeds as well like uh AIS information, no fly zones, uh live telemetry, active and historical findings all into one platform. And we provided the

**[18:38]** customer also with a a dashboard that basically based on some customer workshops we custom developed for them. So they have all they need to report internally on their own uh to their own uh managers and uh and leadership. [snorts] So where we're going with this in the future is one to many operations. So the next step where we see across not just in Singapore but across the world is kind of like okay we need to move away from just one pilot looking after um after after one drone. We need to go one to four in some case 1 to 12 as well already. Uh and that's what we're working towards next. And that's um all I got for you. I'll welcome any questions. Thank you so much Jeff for you know sharing these insights you know of uh giving a you know rundown on the exact value chain right from the solution design to what's being done technically how did you get into the regulatory uh approvals the kind of use cases uh we've already had like a lot of questions in the chat let me see if I can pick a few and we can dive a bit deeper into those

**[19:43]** >> if I can share my screen a little bit longer. By the way, then I can show you the live uh there's a live flight going on at the moment. I can show you the live platform. >> We have we have 9 minutes. So, we have ample amount of time. >> Can you still allow me to share the screen? >> Yes, you should be able to share it. >> Um, not at the moment it seems like. Let me see if I can try again. [snorts] There you go. So you can see one of our drone flights happening here at the moment in the port and you can see here what it's looking at. Okay. Um questions. We're happy for you to to manage your questions. >> All right. So I think the first one is and this is when you closed uh on the last slide. So you're looking at one to many operations and you mentioned that that's the shift that you're seeing across the industry. So like what benefits do you see when it comes to you know shifting from one pilot one drone to one pilot multiple drones

**[20:45]** and uh like how do you plan to make that a success uh for this specific deployment [gasps] >> um basically this would be for for the next the following contract that we're looking to get with with NBA. Um the biggest benefit of course is the cost. uh pilots make up uh a pretty hefty chunk of the overall operational cost of deploying this. Um especially these drones, the the hardware is not the expensive bits. The operational costs are also okay. It's the pilots and the manpower associated with it. [gasps] So that's the main driver in order to get this cheaper and then you can you can scale up further as well. In this case like we would be able to do one to three probably. Uh but what Singapore MPA is looking at is like further like putting more ducks across um across the the coastline as well with this. So I think what we need in Singapore in order to to get to that stage is well we need to to use leverage the new ground control software that you guys have developed which is enabling those multi operations. There's more engagement with the regulatory uh environment where um CAS civil aviation authority of Singapore wants to basically go through our

**[21:49]** overall process and procedures and especially also the emergency process as well. And for Singapore in particular, what we need as well is is access to more airspace so that those return to home routes are not blocked by ships. At the moment, we we we can't see the the drones can't see the ships on the on the way home. So, we need to have a clear path at all times in order to go back and in order to do so, we need more airspace accessible to us. >> All right. So on similar lines you know carrying forward the conversation to BV loss approvals since MPA is a complex environment and you know the screen that you're sharing right now uh does justice to that so we can see you know uh the kind of complex operations you have uh so apart from BV loss approvals were there any specific requirements that you had to cater to get approvals to fly drones and docks again because you highlighted the ceiling altitude ceiling the RTH path. So was there any specific or any specific hurdles that you had to you know cross

**[22:51]** to get to your >> um it's basically a be approval in Singapore not uh not not to too to easily get. So we've been operating Singapore for the last six years. So we have like a proven track record uh with our pilots as well over there. So that that went a long way in order to expedite the approvals uh required for for operating the port. And then um CA gas basically CAS uh is responsible for all air operations and then the board of Singapore was mostly uh interested from like a ground obstacle and location and positioning of the of the docks uh point of point of view. Uh but it was mostly uh cast that need to be needed to be uh convinced with our operations manual, aircraft manual and the process and procedures that we got. >> Right. So I think that's that's what we see in the industry as well. As long as you have a perfect or you know a very well structured concept concept of operations and you can highlight uh all the risks and the mitigations that you'll take uh for your flight operations that's a good enough case to get those approvals with the

**[23:54]** regulatory authority. Uh one question not uh related to the docks but what are the technical and logistical hurdles you encountered in early days when sky ports were trailing the shore to ship delivery. So what do you would like to dive? >> Uh that's done. Okay. Um the hurdles were shipped ashore slightly [laughter] different than this one. So there it was um we did different types of operations with different drones. We did multicopters as well as fixedwing veles. Um I guess the biggest challenge is these um these these ship information like when ships are uh at anchor this information doesn't get updated every second. It gets updated every couple of minutes. Um, and because there's tides that that take place, as soon as you you take off, you're this this data is is outdated already. So, it takes probably 10 15 minutes to fly to the ship, that ship is going to have uh it will have moved. So, you need like updated AIS information along the route uh to update that location. And if you want to land on a ship that doesn't have a tracker

**[24:57]** on board or anything like that, uh custom GPS tracker, that's an additional challenge that you have to deal with. So that's why we moved to multicopter so that the last couple of hundred meters we could take manual control uh and manage the landing especially when we had to operate and land or deliver or pick up parcels in like areas of 3x4 meters uh that's a bigger challenge >> right so since we're already talking about and this is a question from me not from the audience like uh ship to uh show to ship delivery is possibly one use case the current security incident response is one that you're already doing. What other use cases potentially could a port authority have apart from what you've just showed us? >> Um, I think it's um more more just like AI enabled over going forward. So, it's kind of they want to be able to check for example for ghost ships in Singapore. So where they want to see um ships that might have their AIS transponder turned off or they they might have a

**[26:00]** different AIS transponder uh that's broadcasting a different ship name than it's actually the name of the ship. So being able to automatically fly across the port and where the camera auto using AI automatically flags where there's a discrepancy between what the AIS transponder is saying and what the camera is reading on the hull of the ship is something that they're interested in next uh to develop because that's a quite a big challenge in uh where uh based on just the Singapore port location where there's a lot of oil transfers happening between ships in the east and the west uh is happening over there in Singapore. So that's a challenge that they're trying to manage. Um, so I think that's like a example of a of an AI use case. Then we're going to see way more of those going forward where at the moment we have very defined flight paths to look at certain locations. I think AI is going to allow us to just patrol across the area and pick up issues along the way uh and allow us to also then scale up operations so that it's not it it takes a lot less uh pilots per per aircraft in order to monitor these things. So I think that's where we're going forward like it's going to be more of a proper what they're going what

**[27:02]** they're trying to develop is a proper digital twin of the entire port at any given time and this is what what will contribute to that >> right I think one thing that stuck with me was in one of your slides you had mentioned that the output is not just the video feed but it's find you know the findings tied to a location time evidence and that's is that where the digital twin comes into the picture where all of that data is kind of pushed so that it's sort of a living record of all the finding and findings and incidents. >> Yeah, we we developed this because uh MPA is still developing their own digital twin which has like a whole bunch of like a scheduling information as well. So they want to be know how busy the port is going to be in any given time which ships are overstaying their scheduled slot and need to be charged extra. So they want to be able to to see that um and track that. So they there's a whole scheduling element to it as well. So they can forecast how busy the port is going to be, where ships will and which anchorages they they have to be put in. At the moment, a lot of that work

**[28:04]** is being done ad hoc by the pilots going onto the ships and guiding them into the port. So they want to get more control in order to have more more throughput and and maximize their their operational capacity. Um so I think this is something that u we're going to see uh in the future as well. We what we create at the moment is kind of like a dashboard view but like there's basically a whole database attached to this that they could just incorporate into their own system as well. >> All right. And just on the same question but zooming out a bit Mia again like we've established that complex one of the biggest ports. Uh how does such a big customer or a port authority actually embrace such kind of a technology and like did you see any hurdles challenges over there? Uh what was your approach? uh this can be beneficial for a lot of people trying to work with ports or even port managers trying to adopt this technology. So any insights that you'd like to share? >> Um it's what what made it work in Singapore is that you have a couple of champions within the business that are that are that are pushing this forward and and helping us

**[29:06]** navigate such a massive organization like MBA. Like there's like eight different levels of of of management within organization. So finding the right people that can can can help uh be the champion for the service in the business and then once you have the opportunity to deliver like if you then suddenly see like oh we have a live feed of of of a of a containership fire or actually we're able to support um operations where we have to find a missing person that went overboard or we also just yeah what we provide them is um like support on like the innovation days and things like that. So that that that helps people see the benefits and then >> just like don't don't be afraid to just innovate yourself like NPA did not ask for this platform but basically now they can't live without it. So we we developed this on our own because we saw operational benefits for our inter our internal operations and then yeah NPA didn't even knew that they wanted this uh internally but now I think they're very happy that they have it on a daily basis >> right I think those quick wins really matter once you start showing the quick

**[30:08]** wins and the customer starts to see them as well that's when they start to reimagine you know how their operations could be transformed you using drones and docking stations uh we're almost up on time uh Jeff uh thank you so much for uh sharing your story, what Skyore is doing and how you're helping out the MPA. Uh any closing thoughts for the audience over here? >> Uh I think like we're in a really exciting time at the moment like the way that we we see that our our drone use has has changed and the kind of use cases that we're unlocking now, not just in import environments, but construction monitoring is a really interesting one. power line inspections. U there's all these industries are just embracing drone operations way more and and and seeing that like a small investment can yield massive benefits for like the operational uptime. So any industry that has massive penalties for when you're offline like oil and gas or ports then uh drone investments are uh a quick win often that can have massive

**[31:09]** operational impacts and benefits. >> All right, perfect. Thank you so much Jeff for you know sharing those insights. I'm sure I learned a lot. I'm sure the audience learned learned a lot as well. Uh thank you so much for being a part of Neshen and this session and thanks for the audience as well uh for being here. >> Thank you. >> All right everyone.