I’ve been following these as he releases them; just a quick warning that he’s using a $2,500 radar module that’s often backordered. It’s an awesome tutorial but definitely in “serious hobbyist” territory if you want to follow along for real.
Yeah I've been reading about radar recently and this popped up at the perfect time, so I was excited to have a tutorial to building my own... until I saw the price tag. Once I have some time though I'm going to look into the MIT cantenna radar mentioned in the first video
If you're interested in the cantenna radar you might also find Henrik Forsten's radar interesting: https://hforsten.com/6-ghz-frequency-modulated-radar.html. It uses largely the same high-level design, but is a bit more integrated as a single unit (rather than a series of minicircuits components).
I've built the MIT radar before and it is a great way to learn the basics of radar!
In either design you can likely get away with <$500 worth of materials.
Henrik has follow-up blogs adding and modifying this design for more functionality, it's a pretty neat project to follow!
Quick question regarding DYI radars. What's the legality of it? I would assume the radar frequency bands are reserved (I could not find much information on this with a quick google search). I wouldn't want to get a visit from the FCC because my hobby project is interfering with the nearby airport. In the project you linked he is using WiFi bands, but they come with their own restrictions (the very low power limits immediately comes to mind).
Have been using TI’s AWR1443 for mmWave radar development. It’s a bit less pricey than the board mentioned here at ~$300. The software stack is a bit rough though. You can buy another board for $700 to stream raw radar data out for analysis. https://www.ti.com/tool/AWR1443BOOST
I love things like this. The guy is currently on episode 2 / 5 on this radar thing.
Drones have the potential to really become a cool real-life e-sport over time. It'll even be a direct analogue to modern combat, much like early sport was.
There's a market in smaller militaries for stealth hobbyist-level drones. This type of tech is a game changer for military applications that don't have large budgets. I would imagine you start out R&D for such things with this type of radar.
The market is not limited to smaller militaries. As we can see in the Ukraine war both sides are using tens of thousands of cheap 500 dollar fpv drones. Those are by far the most successful new warfare tech introduced in a new war. All the existing expensive drones and also expensiven AA proved to be quite worthless compared to cheap drones.
it is limited to smaller militaries. Mainly because the "bigger military" I'm referring to doesn't purchase commercial products. They purchase from specially vetted and procedure following "defense contractors" that charge a 100x premium.
They'll never be using such a product. Russia is not a "big military" not when China spends 4x as much as them. And China is a small military too, they spend around 4x less then the next biggest military. You'll have to guess what I mean by "smaller militaries" from these details I've provided.
> Mainly because the "bigger military" I'm referring to doesn't purchase commercial products.
Though...the individual solders do bring those products to the battlefield.
I actually got an award during the '03 Iraq invasion for using my personal GPS to rejoin my convoy after breaking-down, getting stuck in the sand and basically committing mutiny when I got tired of my platoon sergeant driving around in circles for some unknown reason. Truth be told I was using a map I pulled out of a burn barrel and a cheap compass but everyone was so happy when we reappeared (we had all the fuel) that I don't think they really cared 'how' and just assumed I was using the GPS.
And it's not like I bought the GPS unit specifically to navigate around in the desert but just brought it along because I already owned it and thought it might come in useful -- which it did on multiple occasions because our platoon leaders loved to get lost, I'd regularly walk up to their humvee to see what kind of navigational shitshow they were involved in and be "yeah, were still 10 clicks away from what you're driving around looking for". Can't really complain too much since they never got lost well enough for anyone to get a mountain in Phoenix renamed after them as we were driving around the exact same roads at the exact same time that incident did happen.
Anyhoo... the US military has a more bottom-up friendly approach where any random Joe can come up with an idea or novel usage of something and it will get incorporated into operational doctrine as opposed to other militaries which are all about a top-down approach to 'innovation'.
If you count artillery units or tank units Russia and China are not small.
One Russian tank costs the same as 20 German ones. In a real war price is what counts. That's one reason why Europe and the US cannot win the Ukraine war. Everything is just too damn expensive. It's not a military show or a bunch of sandal terrorists anymore. It's real.
>One Russian tank costs the same as 20 German ones.
That's just not true. Not sure what tanks are you referencing, but let's take the "flagships" Russian T-90 and a German Leopard 2, which are in a similar category and have similar capabilities. Their basics models both cost roughly the same.
> That's one reason why Europe and the US cannot win the Ukraine war.
We'll we've been fighting this war for over 2 years now. You don't always need a tank to destroy an another tank. To destroy a 5 mln $ tank you could also use a 500$ drone with an RPG round rigged to it.
I think the real game changer is using consumer smartphone grade SoCs for visual based nav+target guidance. The same $500 drone that can navigate through a GNSS denied environment and target a person, that's scarier than manually flying them one by one trying to beat signal jamming. Oh, and they scale better - no operator needed.
I've work on these "secret" systems in anduril. I can tell you all about it, even the secret parts because nobody really gives a shit.
Most companies aren't using SoCs here. They use Nvidia Xavier's and radar. Vision is for record keeping, it's mostly a joke to use it to trigger some counter uas thing. Its like a bird triggering a missile launch. Vision isn't reliable enough... Similar to how like an LLM isn't reliable enough.
All of these systems can easily be overwhelmed by a cheap small cluster of drones. No system will react fast enough to bring down say 10 drones.
A stealth drone will give a solved problem of getting past counter uas with an even cheaper solution. That's all I'm saying here.
Actually a phalanx might have a shot at bringing down 10 drones. But it won't do this consistently and each bullet costs probably around 20 drones.
I’m assuming visual based nav is more sophisticated than terrain following systems like cruise missiles use right? I would imagine these days drones can read and follow road signs to the general area and then loiter while using facial recognition to find their target.
Having specific XP with counter sUAS systems, radar is unlikely to be a working solution given the range, speed, size, and cross sections were talking about for a realistic deployment.
Fun idea but likely not going to handle the the claimed use case.
> without an array of camera systems in some indoor environment, drone research is a non-starter right?
No, that is not right. For practical reasons many people working on control algorithms for multicopters do use facilities like you describe.
The indoor nature helps managing risk to third parties and decouples your experiments from the weather. Many experiments are helped by an external ground truth localisation data, and camera based systems are relatively cheap and relatively simple way to achieve that.
That doesn’t mean that multicopter flight control is the only thing worth researching about drones. Nor does it mean that every possible experiment requires a sub-centimeter external pose tracking solution.
