# I Built a $40,000 Military Drone for $120.07 ## Метаданные - **Канал:** Data Slayer - **YouTube:** https://www.youtube.com/watch?v=bmLE9BT76Pc ## Содержание ### [0:00](https://www.youtube.com/watch?v=bmLE9BT76Pc) Segment 1 (00:00 - 05:00) A defensegrade recon drone runs taxpayers about $40,000, but this build costs $120 and a Saturday. The government's probably going to hate this. And honestly, it's starting to look like a pattern. First, I built the Houdini, a $999 GPS dog collar rebuilt for about 60 bucks, no subscriptions. Then came the Haven, a $20,000 military mainet radio hacked together for 106 bucks. And now a $40,000 surveillance drone reverse engineered in my office for $100 and change. It's almost comical how much markup these big guys get away with. After the last minute build went semiviral, my inbox lit up. Crews from the World Racing League were looking for better pushto talk out on off-grid racetracks. DARPA contractors trying to leverage this tech. Apparently, you don't need a program office to ship slick tech. But we already knew that, didn't we? But building the network was just the beginning. Now the question is, what can we actually do with it? If act one was replacing $20,000 radios with Raspberry Pies, then Act 2 is building surveillance drones from scratch. Why? Companies like Andrew are raising billions right now to build drones so advanced they barely resemble anything we've seen before. But here's the thing. For the cost of a tank of gas, I built one in my office that can stream live video into tech. the same tactical software used by the military. And I can do all of this without being at the mercy of the FAA, DJI, or big defense budgets, which sounds like more fun to me than building the next Loheed Martin. Sure, you can always throw money at problems. I mean, defense contractors do it every day on the taxpayers's dime. But what actually impresses me isn't a blank check. It's the resourceful builders who find ways to get more with less. That to me is real progress. The problem is most of us don't even own our tech anymore. It's locked behind subscriptions, DRM, and cloud accounts that we never asked for. Our devices spy on us, and every update seems to push us towards yet another monthly fee. One viewer told me a story that really stuck. Back in 2010, he helped launch a project called Green Canopy. They took access points headed for e-waste, flashed them with open WRT, ran them on solar batteries, and meshed them to provide free Wi-Fi across Nashville, Tennessee. It also doubled as an afterchool program where high schoolers learned networking, electronics, even business skills. But once it started gaining traction, big tech lobbyists went as far as passing a bill banning municipalowned networks. Stay classy, big telco. And it hasn't stopped there. Now you see companies like Google killing off sideloaded apps, tightening the leash even more. With every passing day, we're sliding from an ownership model to a licensing model. Which is why I'm so passionate about building a parallel tech stack, open-source, inexpensive, capable hardware we can actually control so we can call the shots, skip the red tape, and ditch the bloated monthly subscriptions. Now, I've been testing decentralized wireless communication systems for a few years now with one goal. Put the power back in the hands of the people. And that's what this channel is all about. Helping you deploy open, resilient systems that cut out the middleman and give you control. All right, stepping off my soap box now. Let's see if this actually works. Today, I'm strapping a $6 Halo enabled chow board with a micro FPV camera, what I'm calling Scout, onto a drone. The mission is simple. Stream live video into ATAC, maybe even to YouTube live. Sounds insane, but here's the kicker. If it works, it's something billiondoll defense drones still can't do. Those systems are locked down, proprietary, and stuck behind program offices. They don't talk to civilian apps, and they don't push straight to the open internet. And you definitely can't mod them in your garage. With Scout, you can because when you build it yourself, you're not bound by contracts. gatekeepers or firmware limits. You can strap it to a $100 drone, plug it into a mesh, and send a live feed anywhere in the world. Here's the plan. Step one, create a mobile ad hoc network in a signal dead zone. Step two, drop in a Starlink uplink to give the network access to the wider internet. Finally, step three, connect our drone mounted chow, code name scout, to the mesh and try to stream from the air. And to know if it works, we're running three field tests. Test one. Can we stream realtime FPV directly into ATAC? This is going to be raw overtheair video fed straight into the same tactical mapping software used by the military. Test two, will the Halo Link survive at altitude? Most consumer signals die after a few dozen feet. If Scout can hold at 60 ft, then we've got something special. Test three ### [5:00](https://www.youtube.com/watch?v=bmLE9BT76Pc&t=300s) Segment 2 (05:00 - 10:00) can we punch out to the open internet? If we've got ATAC working, the next leap is global. YouTube live, remote surveillance, or relays to anywhere the mission needs. But building the tech is only going to be half the story because there's a second battle. Staying compliant. If you're flying DIY hardware, you've got to think about everything. FAA ceiling limits, line of sight rules, even the weight of your rig, and whether it pushes you into remote ID territory. That's part of the challenge here. Can Scout not only deliver the range and resilience of a $40,000 defense drone, but do it while staying inside the lines? Hacking the tech is one thing, but proving it works and keeping it legal, that's the real test. Now, the backdrop here makes it even more interesting because earlier this year, the US government moved to blacklist DJI again, threatening a nationwide ban on their drones over security concerns. DJI fired back by quietly lifting most of the operator restrictions baked into their firmware. That means no more no-fly zones, no forced altitude limits, and no automatic grounding. Their drones can now fly just about anywhere with almost zero guard rails, which technically bolsters our cause, but also puts the onus on the drone operators to be smart about what they're doing in the sky. Now, outfitting a drone with meshtastic or a halo node might sound like a novelty, but it's actually deeply practical because both of these technologies rely on line of sight. So, trees, hills, and buildings block signals. And when that happens, the mesh can break down. But lift one node into the air, even briefly, and the whole picture changes. Suddenly, you've got an ephemeral cell tower in the sky. The elevated link restores coverage, stitches the network back together, and extends range far beyond what ground nodes can do. And if you got a Heltech Laura 32V3 or the newer V4, I've built a cheap adapter that'll outfit it onto your drone. Links down below. Picture a hiking team losing GPS because a canyon wall blocked their signal. launch a drone for a few minutes, and that airborne node reconnects the mesh, reacquires the missing positions, and guides everyone back to base. At least that's a theory, but practice is always another story. So, let's stop pontificating and get it built. Time to fly. Now, life-saving tech shouldn't be an upsell. Big brands charge subscriptions just to send a message off-rid, and that doesn't really sit right with me. Staying connected when it matters most shouldn't depend on your wallet or on whether a corporation flips a switch. That's why I'm building open decentralized tools you can buy once, run as long as you want, and even repair or teach your kids how to improve. Now, this drone is old faithful, my battleworn DJI Mini SE. It's one of the smallest fullfeatured drones that DJI makes. stock, it comes in under 250 g, which means it skirts most FAA rules. But once I start strapping payloads on the FPV camera, the Halo node, I'm pushing past that threshold, and that changes everything. Old Faithful's been through it all. Crashed trees, bounced off pavement, baked in the Florida humidity, and somehow still flies. She's not flashy, but she's tough. And today, she's got a new mission. Carry Scout into the sky. The catch? Drones over 250 g are legally required to broadcast something called remote ID. Basically, a digital license that beams out your drone's position, your location as the operator, and its serial number to anyone nearby with a receiver. Now, this isn't the same as ADSB, which man aircraft use to broadcast their position to air traffic controller and other planes for collision avoidance. And you've probably seen people online geeking out over ADSB data, tracking everything from commercial airliners to military transports. By law, every manned aircraft in the US has to broadcast it. Private jets, cargo haulers, even the planes of visiting heads of state. It's all out there. Public data. Remote ID isn't quite that. It's a shorter range, ground level, and really just a local accountability beacon. more like Bluetooth than aviation radar. A digital license plate, not an air traffic system. To me, it feels like one of those hard to enforce laws, and I don't see much evidence of law enforcement actually checking for it, but technically it's still required. Newer drones have remote ID baked in, but older and lighter models like the Mini SE don't, which means you either shed weight to stay under 250 g, or you strap on a third party remote ID module like this. The RUO R111S, FAA compliant, lightweight, runs about three hours on a charge, not super elegant, but it keeps me legal, and lets me fly this mission without any gray ### [10:00](https://www.youtube.com/watch?v=bmLE9BT76Pc&t=600s) Segment 3 (10:00 - 15:00) zones. Now, while this is my first time putting a Halo node in the air, I've already done plenty of tests flying meshtastic nodes. A friend and I once managed to text over an encrypted channel pointtooint more than 5 miles in the sky. A few weeks later, we pushed harder, trying to stretch a connection from Jupiter, Florida, all the way down to Miami Gardens, which is about 60 miles. But we didn't quite make it. Even with the long, slow preset and some serious altitude, his drone was at 300 m and mine was at 120, the link didn't hold between Fresno clearance, earth curvature, and the limits of my altitude. It just wasn't happening. But I did later manage to lift my altitude cap to 500 m, which is kind of insane. So, next time we'll try again with more height and stronger radios. Speaking of stronger radios, I recently got my hands on the new Heltech Laura 32V4, which pushes the dBm from 21 to 27. And in terms of raw power, that's about 125 m to 500 m, which is a four times jump in power. And this isn't even on the shelves yet. It basically just showed up at my apartment in a very innocuous box, but that's going to be a whole separate video. Today is about Halo. On the ground, I've got the Haven mainet tied into Starlink. That's our network core. It's handling both Halo and traditional Wi-Fi, which my Android phone uses to run ATAC. If you want to build one yourself, I've got a full guide linked below. And in the air, that's Scout, a stripped down Halo node running a $6 Chia MCU from Seed Studio. It can't do fancy meshing, but it can spin up an HTTP camera stream on a web server and make it available to anything else on the network, and that's more than enough for this test. Let's break down Scout because honestly, it's shockingly simple. The Brain is a $6 Chia MCU from Seed Studio. Basically, a tiny ESP32 board. The eyes is a barebones FPV camera, just raw video, no frrills. The link is a Halo Wi-Fi hat to push IP packets over long distance. And the power is a single 18650 battery. Total overkill, but works in this case. I like this one because the side USBC port makes powering small boards dead simple. And for the code, custom firmware streaming a 1280 x 1224 SXGA at about two frames per second tuned for ATAC. You can definitely push higher, but frames start to stutter. I did try pulling power straight from the DJI's USBC, but no dice because it doesn't seem to kick out any power. So, for now, Scout runs on its own pack. Not elegant, but that's the point. It's a proof of concept. And if you've ever wondered how I'm able to turn ideas like this into working firmware so quickly, I put together a short microcontroller prototyping guide linked below. And it just walks you through how I use secure LLMs to rapidly prototype, iterate, and test firmware builds without wasting weeks on boilerplate or deadends. Now, once I was in the field, I spun up the Starlink uplink, fired up the Haven node, and strapped Scout onto the drone. The rig wasn't pretty, but it was enough to test. Down the road, I'll build something cleaner like the Sky Mesh mounts I've been using for Meshtastic. And with that, it was time to fly. Okay, so we have our dual mainet set up uh with an uplink, which is a Starlink. We have a connection. Uh one of the main nets is going to be this guy right there. You can see it's just running independently uh because of those big batteries we have on there. And then the other one is Etherneted into the uh Starlink router. And then we have the Starlink right here. And now we're going to try to connect the cow and get a live stream down to ATAC from the drone. So we'll see how that works. Number doo. Doohoo. ### [15:00](https://www.youtube.com/watch?v=bmLE9BT76Pc&t=900s) Segment 4 (15:00 - 16:00) Before we get to the results, I want to point out something I noticed while messing with the DJI Fly app, RTMP streaming. DJI lets you push your live feed straight to YouTube or Facebook, which is pretty slick, right? The problem is it only works if your controller has internet. No bars, no stream. What we just built is kind of the opposite. Halo streams into the mesh first. It works in dead zones and only uplinks if you give it Starlink or back hall. DJI is cloud first. Ours is mesh first. And that's the point of our builds. It works even when the grid doesn't. And it worked. Scout held the link. Live video streamed into ATAC 60 ft up. Of course, interference was a real issue here because we had multiple radios fighting in close quarters. The feed showed weird ripple waves, and the propeller on the drone even sliced straight through the Halo SMA connector. But the crazy thing was it still worked. $40,000 versus $120. 7. Keep the extra $39,879. 93 for, I don't know, fixing potholes. For more killer tech, click --- *Источник: https://ekstraktznaniy.ru/video/29819*