# How DJI Dock 3 Is Powering Autonomous Operations at Scale Worldwide - Event: NestGen ’26 - Published: October 2, 2026 - Length: 30:04 - Watch: https://nestgen.org/library/watch/-N_1q3nFRp4/ > 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 Scaling autonomous drone operations from a single pilot site to a global fleet is one of the hardest problems in Physical AI. Teams need docking infrastructure that performs reliably across climates, integrates with existing command-and-control systems, and keeps getting more capable without adding operational overhead. In this NestGen live case study session, DJI Enterprise will walk through how DJI Dock 3 is powering autonomous drone-in-a-box operations at scale for customers worldwide, and what it takes to get that infrastructure production-ready. ## Transcript **[0:00]** into physical AI more broadly looking at systems that you can deploy and put them to work for various use cases. Uh during this uh negen we're going to have a great mix of people building and deploying this this stuff. So I think we're going to have some really great conversations. Now let's talk a little bit about this uh specific session. Uh it's about it's the physical AI showcase and one of the most interesting things about physical AI is that we are moving from AI that can perceive and reason to systems that you can actually go out and deploy in the field. Autonomous drones and docking stations are a pretty good example of that. But getting one autonomous system to work is one thing. Getting it to work reliably across different sites, different use cases, different environments, different weather conditions is a very different challenge. **[1:02]** And that's exactly what today's session is about. How DJI Dog 3 is powering autonomous operations at scale worldwide. For those of you who I haven't met, my name is Neil and I head customer success at FlytBase and I'll be moderating to today's session. So before we get started, just I'd like to plant a seed um you know uh so that you can have that question in your mind. So if you're starting your drone program or you're kind of in the middle of it or even figuring out how you want to scale your drone program, think about what these autonomous systems can do and how they can reduce the work of humans out there in the field and create a lot of value. Um the audience who've joined us please let us know where you're joining us from you know uh tell us again which stage of your journey you are in. And without further ado I'd like to introduce Molly Huang drone in a box specialist uh at DJI who works **[2:06]** with the global product management team as well. Uh Molly welcome to the session. >> Hello. Hello guys. >> Right. So before we get into the session just a couple of you know housekeeping rules as well. So the session is being recorded. Uh we will definitely have uh a Q&A session. So you know if you have questions popping up in your mind mind you can just put them in the uh Q&A session and we'll take them up uh at the end uh of the webinar as well. And yes without any further ado let's go ahead. Uh let me just start. All right, Molly, over audio. >> Good. >> Hello everyone. Uh, my name is Molly. Um, and **[3:12]** I'm a drone in a box specialist with DJI Enterprise. So today I want to talk about how DJI do 3 is transforming the remote operations covering routine inspections as well as time-sensitive emergency response for public and private safety issues. Next page please. Yeah. So here's the plan for today. So first I will quickly review some key features of DOC 3 and then we will look at two real world cases. The first is solar hotspots inspection and uh second one is drone as uh first responder DFR. So they are very different industries but they rely on the same platform strengths. So the automation all day availability, consistent data and fast access to information and then I will leave uh five about five minutes for Q&A. Uh next page please. So, uh, let's talk about the hardware first. Doc 3 is designed to, uh, stay outside **[4:15]** and so it features, uh, significantly upgraded hardware protection. DOC 3 carries a IP56 rating while the Matrix 4D and 4TD aircraft are rated IP55. So, it can operate in standard uh, rain conditions. And it als it operates uh from minus30 uh degrees Celsius cold to 50° C heat uh ensuring reliable performance in both summer and winter and even in strong winds uh with up to 12 meters per second. It can also remain airborne for up to 32 minutes. Uh next page please. So a single drone can fly forward for up to uh 54 minutes with uh up to 45 uh 47 minutes maximum hovering time and the dock can charge the battery from uh 15% to 95% in just about 27 minutes. Uh so this combination supports frequent missions **[5:20]** with turn less turnaround times. So it also gives the operators more flexibility. Um so uh the operator can cover a larger site, spend more time observing a target and keep a safer reserve without sacrificing the useful part of the task. Uh so now let's move to our first use case. Uh next page please. >> Yeah. >> Sorry, I think you just stopped sharing. >> Just hold on. Yeah. Meanwhile, I'll just introduce a bit about our first use case. Um the solar hotspot inspection. >> So when we >> Yes. Yes. >> So when a solar farm uh when we talk about the solar farm, it is more than uh uh just panels. It includes uh inverters, electrical systems, batteries, substations **[6:21]** and the supporting infrastructure. With drones, we normally combine two inspection types. Thermal inspection to find hot spots and underperforming modules and visual inspection to check vegetation, racking, wiring, roads, and physical damage. Uh next page, please. So why do we use a dock instead of a traditional drone program? Um uh the first advantage is automation. Uh the dog enables fully automated takeoff, landing, charging and mission execution without a pilot on site. Second is its all day availability. The tasks can run on schedule or on demand. Third is the consistency. uh the same route and same camera angles will produce comparable data every time. And finally uh efficiency uh so one remote team can support more sites with less travel time and preparation. **[7:24]** Uh traditional inspection often depend on who is flying, how much time is available and when someone can travel. The doc basically removes much of that variation. So the business outcome here is quite clear. It's more coverage, better consistency and less downtime. Uh, next please. So, I will spend some uh next please. >> Yeah. >> So, I will spend some time introducing the dog three workflow for this case. So, the first step to deploy a dog uh we need to confirm the power uh network connectivity and ensure enough open space for the safe takeoff and landing. So connectivity can be hardwired Ethernet, Starlink or another suitable link. Getting this foundation right is quite important. And once it is in place, the host system is ready to go and it no longer requires a new uh deployment every time. Next, please. **[8:27]** Yeah. And before uh we go into detailed inspection, we normally create an updated map or a 3D model of the site. A satellite image is useful, but it may be old. So a mapping flight can give us a current view of the site including uh recent constructions, terrain, obstacles, and asset uh locations. And that will create a better planning surface for future missions. It also gives the operators and the maintenance teams a one shared reference. So everyone now is looking at the same current site model rather than working from different maps and assumptions. Next please and can you click again? Yeah. Yeah. So solar sites are unlikely to be flat. uh they may contain heels, BMS and changes uh in panel uh or or changes in panel elevation. So this terrain **[9:29]** following feature allows the aircraft to maintain a more consistent height above the ground. Um the operational benefit here is that missions will remain repeatable across the full site even when the terrain changes and the technical benefit here is that uh more stable image output uh and thermal quality. So again the selling point here is not the terrain following feature itself but uh it's consistent data without requiring the pilot to manually compensate for every change in elevations. Next, please. So, the third step is mission plan. I'm not going to go into too much details here, but at a high level, the value is uh the DJI uh the DOC 3 can standardize the camera's angle uh coverage and positioning across every task. So, producing repeatable and decision ready data. And with consistent capture and RTK precision, **[10:30]** teams can compare the same assets over time, locate anomalies quickly, and send maintenance crews to the right place with more confidence. Um, so the goal here is not just to fly efficiently, but to turn every flight into some reliable evidence for future action. Next, please. And this is what a real inspection route can look like. Uh you can see uh the structured uh flight lines aligned with the panel rows. So when the drone follows the same route every week or every month, teams can monitor the changes over time. And we also have the change detection feature. So they can uh see whether a hotspot is stable, getting worse or has disappeared after the maintenance. This turns the inspection from one-time snapshot into a continuous view of the whole asset house. Um, next one please. **[11:31]** So, um, so um I'm just going to keep going but I cannot see the screen. Sorry. >> Let me try again. Molly, just a minute. >> Sure. Yeah. So during each flight uh both RGB and thermal data can be captured together using the matrix 4D uh uh aircraft and the thermal data uh shows where the abnormal temperature pattern is and while the RGB provides the context behind it such as dirt, debris or physical damage and combining both field reduces unnecessary follow-up visits and uh gives the maintenance team stronger evidence for prioritizing the repairs. Next one please. And um so uh flight hub 2 uh will brings the operation together. The the system can handles **[12:37]** the flight uh plant flight while the operator monitors the telemetry coverage and live camera output remotely. So our our goal is hands off by default but not hands off without awareness. So the operator can still remain informed and can take control whenever uh they sense any condition changes. So this gives the users the efficiency of automation while keeping the human judgment in the loop. Uh yes uh next page please. The auto dep dispatch uh feature uh moves the system from scheduled automation to a eventdriven automation. An external alert such as a scattera event can trigger a predefined mission without waiting for the next schedule inspection. Uh the operator can define the trigger define the flight parameters and the required action in advance. **[13:37]** uh FlyHub 2 can then assign the available dog and send the aircraft to verify this situation. And this does not remove human oversight, but it removes the delay from receiving an alert to getting useful eyes on the location. And this video just shows that the operator can define the parameters and the action that he he wants it to be. Uh next page please. And here a POI uh which stands for point of interest workflow is another example. So when a specific point needs attention the aircraft can flies to that location and performs a targeted action such as an orbit or focused data capture. These details are configurable but the business value here is simple. uh detect issue, rapid verification and a consistent record without rebuilding the flight **[14:41]** every time. Yeah, next page please. And here are just the typical output we are looking for uh on this solar inspection case such as uh soiling cracks, damaged components or thermal signatures such as hotspots. Um so basically we have a platform collecting evidence that supports uh the maintenance decisions prioritizations and uh performance reporting. Next page please. So after captures all the med uh all the photos and videos the RGB and thermal data can move automatically into an analysis platform or into the customer's own storage and anal um or a stack through FlyHub 2. think um this removes manual download steps and the standardized the handoff from flight operations to analysis. So we also support uh different data strategies include cloud deployment, private **[15:46]** storage or uh more controlled environment where data ownership is especially important. So the complete solar value proposition is not simply just automated flying but it's automated collection of photos, consistent evidence and faster maintenance decisions. >> Uh next please. Next please. And now we will switch to um switch from scheduled inspections to urgent response. Next please. This slide shows why um the 10second uh takeoff matters. Uh the updated logic shortens the selfch check process, keeps the GNSS ready and supports the emergency launch without unnecessary waiting. So for inspection, this can save time, but for public or private safety issues, it can change the outcome of the incident. **[16:48]** Next one please. >> Yeah. So for for public safety issues, they normally face several common challenges. The first is incomplete incident information from the initial call. And the second is limited manpower. The third is travel time. So incidents uh incidents may be spread across a large area and officers cannot be everywhere at once. A drone can arrive first and provide a life wheel before the ground units come uh arrive. So it will it can help confirm whether a threat is real. Uh assess the urgency and even clear an alarm when nobody is present there. So this is a major shift. Um, the doc turns the area awareness from an oa occasionally occasional capability into an ondemand service. So instead of asking um, do we have a drone or a flight of a pilot available, the **[17:54]** team can ask which dock is the closest one. Yeah. Next, please. So I want to highlight uh four capabilities that stands out for this DFR. First is flight to locations can guess the nearest suitable aircraft move to the uh target location quickly and the virtual cockpit gives the operator control of the aircraft and payload. The autod dispatch uh feature can connect external alerts such as CAD or sensor events to predefined missions and flight safety combines uh geo fencing aircraft alerts weather weather information and monitor conditions. So each of them solves a different part of the problem response response time decision quality integration and operational risk. uh together they create a system that that is fast, useful and manageable from a centralized command center. **[18:59]** Next please. So I want to highlight this flight to location uh feature because it is designed to be quite simple and user friendly and it can also uh be done on the DJI enterprise app. Um, so you simply user just simply search for an address or click the location on the map and the closest suitable dog can launch within 10 seconds. Um, and a flight 2 can also be issued when the drone is already in the air. So the value here is very immediate. The eyes on the scene before most of the ground resources can arrive. And that really gives the dispatchers and the teams more time to choose the right response rather than making the decision using only a phone call or a map p. So here we are not trying to replace the responder but we are helping the responder to arrived with better informed um information. **[19:59]** Next please. And once on the scene, the virtual cockpit uh supports a dynamic response. The operator can switch between wide zoom or thermal wheel uh views. Uh you uh use the laser rangefinder, add the spotlight or add a speaker, and take manual control when conditions change. And the AR overlays can show the street names, businesses, and house numbers directly on the uh uh on the screen. And the live stream can also be shared through a web link. And the our latest DJI Enterprise app also support a live stream. And it can also share your live screen through the app to your group members. Next, please. And one more thing worth highlighting is the AI powered uh object realtime int uh **[20:59]** intelligent recognition. Once enabled um the detected objects are display are displaced in real time on the live stream. The algorithm supports recognizing people, vehicles and boats. So here's a quick story uh about this uh real-time recognition. on one team's on one team's uh their first uh DFR flight with Doc 3, they are actually searching for a person and the AI flagged someone lying on a rooftop >> rooftop. So honestly uh uh the people would never have spotted that with the naked eye and I want to say that's the where the uh the point of these detection features they can act as an extra pair of eyes for the operator. so they can uh make faster decisions. Next one please. **[22:00]** >> And this uh this is a a real world case um ash deployment shows that uh what it looks like in practice. Um and there's an also a like a a YouTube video for this. If you are interested you can search for that. For the time concern, I'm not going to play it today. Um, so after the hurricane Helen, a private uh uh a public private partnership helped to bring the DFR capability to this region and in total 13 docks were deployed. When roads and flooded areas were difficult to reach, drones provided the overwatch, helped direct officers, and supported the recovery effort. And the same network also flew repeatable mapping uh routes to document cleanup and infrastructure conditions. An agency could also re uh revisit the same locations and share updated imagery **[23:01]** without sending additional crews into those unsafe areas. So uh this shows that a doc can scale from immediate response into recovery mapping documentation and long-term coordination. So across both these uh use cases the same capabilities um keep appearing. It's the in upgraded hardware, the remote operation, the reliable communication, uh repeatable data automation and integration with other systems. And um and in both of the cases, uh people can still remain in control. They simply just spend less time on repetitive setup and more time on acting on useful informations. So the industry are quite different but the overall uh fundamental operating model is the same **[24:02]** and that's I think that's it for for today's uh sharing um I will leave some time for questions any questions >> well thank you so much Molly really interesting insights u you know and I think I I can see a couple of themes on you know uh how uh a reliable hardware such as the DJI doc can be deployed for multiple use cases and solve multiple problems for a customer specific deployment. Uh so let's let's dive I think we have four minutes let's dive a little bit more uh into the questions. Uh okay the first one it's interesting when is the DJI doc 4 launching? I think everyone is waiting for that. >> Yeah, we are excited. Um I I think it'll be pretty soon, so we can just count the **[25:02]** days. I'm not allowed to disclose anything, but just wait for it. [laughter] >> All right. I think I think we're all excited. Everyone is is excited for it, and we can't wait for the newer version of the doc. Uh okay. One question. Okay. So uh the idea is like if you compare the dog deployments from like today to two years ago uh the most of the deployments were like single dog uh pilot programs but now like let's say in 2026 what change are you seeing in the deployments when it comes to the number of docs that people are deploying and managing at the same time? Um I think uh compared to two years ago uh where only single site can be deployed now we support um multiple sites deployment and multiple joint deployment. So the advantage is that **[26:04]** we enlarge the total area that we can monitor and um also uh additionally two drones can work together. So one drone can work as a relay for another drone to fly to a more extended area. So this basically enlarge the uh area that uh the doc can reach to and that is a very uh I would say exciting improvement to into real world practices. >> All right, perfect. So since we're talking about multi-sight deployments, I had one question and there's a similar question uh from the audience as well. So across all the global deployments uh that DJI has done which is by far the one use case where you're seeing a lot of traction. So that's the first question and the second is which is also the most surprising use case that you've seen a customer deploy a docking station for? **[27:06]** Okay, I'm going to think about that one. Uh, so sorry, what's the first question again? Can you can you mention the first question was across all the DJI doc deployments, Doc 3 deployments, which is one industry that has adopted a lot of docking stations. Oh, I I think it varies by regions because in in Australia or in Europe or in Asia, they all have different scenarios that need a dog. So, it really depends on the regions. Uh for example, in Australia, it's most of uh mostly mining scenarios and >> uh in in Europe, it might be DFR, private safety. And in China here it might be uh also private safety construction uh and uh substation and so on. >> Yeah. But overall I would say substation um the solar inspection would be the most common use cases. And the most surprising one uh I I would say is the Asheville cases because I was **[28:12]** always like touched by uh how it really helped the area goes into recovery and help people in in danger or in unsafe area. Help them to get more information uh get more um uh like food or necessary needs that they need on time. Right. I think that's the beauty about drone as a first responder use case, right? You can deploy it for humanitarian efforts and that's that's the really uh that's a really great part about it. So I think Molly, we're almost up on time. Uh first of all, thank you so much for you know uh coming online and sharing the different use cases, the stories uh and your insights. I think the common theme is how can we make sure that we can consistently and automatically and you know capture data and let respond to multiple use cases and incidents. **[29:14]** Uh any final thoughts on that you'd like to share with the audience? >> Well, I would just want to say thank you very much for the time and joining this webinar. Um please uh do focus on the doc for launch. Yeah. >> All right. Thank you so much Molly and thank you so much for the to the audience as well uh for you know being engaged and joining this session. Uh we have a lot of more sessions lined up. So looking forward to seeing you there as well. Uh thank you. Bye-bye. >> Thank you. Bye-bye.