Fastest camera drone vs. F1 car (ft. Max Verstappen) [video]
youtube.com
youtube.com
See various indie Steam games. 90% of them go unnoticed by the general public despite being decent and fun, but sometimes some famous streamer picks one up and then that one game explodes in popularity despite many other being better than it.
It's just how life works. It's full of randomness and brownian movement.
https://en.wikipedia.org/wiki/Mona_Lisa#Refuge,_theft,_and_v...
Wondering why this had to be manually piloted versus automatically tracking, possibly getting direct feedback from the vehicle about braking and accelerating and turning, so that the video would be smoother and it could fly closer to the vehicle?
I didn’t find the piloting to yield a particularly smooth video, always reacting slowly and inconsistent with its distance and angle.
Lack of skills, time and resources for a prototype project on such a short notice and tight deadline?
A team of 3 drone enthusiasts don't have DJIs resources to deliver on-device object tracking computer vison for the exact time Red Bull gives you public access to their new F1 car, considering they were struggling to even get the drone airborne and fly properly in the first place.
We also simulated driver profiles, so you could adjust them if they wont give you the real data.
Rockets have an even more predictable trajectory and yet NASA with all its resources can’t create an autonomous tracker that isn’t filled with all sorts of stiff, robotic movements and corrections.
I thought the filming was great under the circumstances. And the mix of wider and close shots was very informative and interesting. Just because a more perfect outcome can be imagined doesn’t imply it’s actually better.
Nice flex but your argument is unconvincing. Yes the track layout is well known and predictable, but you only get a few GOs with the F1 car on that track and there's a million ways your untested flight and computer vision tracking algos you set up in 3 days can backfire spectacularly, so why not stick to manual flight from a seasoned drone pilot who can handle all the edge cases?
Automatic tracking of any kind in drones is still not a commodity; most DIY FPV drones like these use flight control software which is ridiculously primitive by control systems standards, just PID rate control with some aggressive gyro filtering (PT1/2/3 and dynamic notch following motor RPM).
There's a lot of academic research in this space - GitHub is littered with thesis projects about drone autonomy of all sorts (inside-out, outside-in, inertial based, visual based, basically anything you can imagine), and there's a lot of commercial success (DJI, Skydio, Autel, and Parrot all have decent inside-out visual odometry and follow-me capabilities), but no project has yet emerged that bridges the gap from research one-off to commodity download for custom purposes.
Yeah and even there, those commercial automatic tracking systems are only meant to follow you mountain biking at 30 kmph, not driving at 10x that speed in your F1 car.
F1 themselves use cameras mounted on cables above the track traveling at high speeds for tracking shots. Having drones fly at that speed and accuracy would be a monumental challenge, not to mention super dangerous and expensive if one were to fall down at that speed and hit a car/viewer. It could probably kill someone easily. Hence my point of these having military applications.
Considering the context, I thought it as pretty impressive. I'm sure if it were for a feature film it could have been a bit better in aesthetic composition and framing but a movie would have had a lot of rehearsals, section-by-section planned coordination and also not have been shooting a full lap as one continuous shot. Given that this was a Red Bull marketing video vs a movie, I suspect they didn't have Max for very long (hence having to go ahead with the shoot despite the lousy weather). IIRC, there are even F1 rules strictly limiting the number of laps an F1 team driver can do in an F1 car outside of officially sanctioned practice sessions.
Interestingly, we'll soon get to see a feature film interpretation of F1 racing. A major movie starring Brad Pitt was shown shooting around the pits last F1 season and Brad Pitt was interviewed about it on one of the F1 pre-race shows. I expect we'll see some dramatic aerial shots when it comes out.
As a side note, these kind of drone shots are often done with a wide angle lens which gives some extra leeway for image stabilization and re-framing in post-production. I expect that was probably the case here because I don't think there was enough room in that bullet-shaped drone for a camera gimbal.
That this is a damn guided missile. I want one.
So using relatively inexpensive parts they built an artillery shell shaped drone that can travel at over 300 kmph ... for just one lap.
Do you hear that? It's the sound of every military in the world ringing their phones.
For context, commercial DJI drones are heavily used in Ukraine right now by both sides mostly for recon (and more), but since they're slow, they often get shot down by rifle/small arms fire form soldiers. However, if you could make them go 300+ kmph, even John Wick couldn't shoot them down while they do a recon lap.
Using multi milion euro tracked vehicles firing expensive 35 × 228 mm ammo to shoot down drones the size of a Starbucks Venti costing less than and Apple Vision Pro is like using your AR-15 to shoot mosquitos. While I'm sure someone in Florida has already attempted that, it's the definition of asymmetric warfare.
Yes. The Royal Navy has used £1.2M Sea Viper missiles to shoot down drones in the recent Red Sea conflict. They have also used their Phalanx guns to shoot down drones, which is much cheaper than the Sea Viper.
> it's the definition of asymmetric warfare.
Totally. My point though was that it is possible for modern counter measures to shoot down fast moving drones, even if John Wickes can't. IIRC Phalanx was used to shoot down incoming mortar rounds for base protection in Afghanistan.
The travel distance for #9.5 birdshot is only 700’ - and that’s ground distance covered when fired at optimal angle, not vertical reach.
I've seen videos of Russians throwing rocks or magazines at drones, they'd have better odds with shotguns and birdshots
I presume weight is heavily optimized to get 300kmph. So it won’t go that fast if it carries explosives. Could be used for high value target killing. Like literally a bullet you can control, might be able to make that work.
But for reconnaissance can’t the drone just go high up ? And then can’t really be shot down with a rifle ? You just need better optics.
Please stop speedrunning scifi dystopias
"The phone of Thomas de Koster from Middelburg has been ringing off the hook since the video came online yesterday afternoon."
https://www.omroepzeeland.nl/nieuws/16315140/snelle-drone-va... (Dutch)
The modern battlefield is getting even more scary. Soldiers are gonna have PTSD from the drone motor whine when they return home.
If you have enough of them you could use them to sweep out areas. Launch a bunch and have them map out the place. You can have them carry weapons and other such things, eg laser communications networks.
Back then I figured that by the 2020-2030s ships would exist that would primarily carry (larger) drones. Those drones could then carry these smaller drones into their locations to do all the things they need to do.
- Torbjörn, Overwatch
For example this rather depressing article cites Russian manufacturing capacity as outstripping Ukraine: https://www.foreignaffairs.com/ukraine/ukraine-losing-drone-...
As per the speed, on their way down current (very cheap) drones are already way too fast for anyone to try and shoot them, the taking down (when it rarely happens) usually takes place when the drone is in the middle of a hovering pattern. Were the drones to always go at full speed (100 kph, 300 kph, it doesn’t really matter) then the operators controlling them would have no way to tell which is which on the ground.
Do they? I have an old and basic Mavic Air, but I can't imagine the skill it would require to shoot it down with a rifle if it's moving high up in the air.
It's difficult enough to even spot it.
I'm sure they do get shot down sometimes, but 'often' enough for it to be a problem?
Certainly hunters have had enough luck shooting ducks and other waterfowl with similar speed. Albeit with shotguns.
For one person with a bolt-action rifle, no.
For a battalion of conscripts with automatic rifles? I’m sure it happens quite a bit.
Building this stuff is trivial too. Anyone who can solder a few components and follow a 30 minute youtube tutorial can have one of these.
Another problem of multicopters is the loud noise they produce, not really suitable for warfare.
I think there is a reason that most military drones are fixed-wing aircrafts, similar to regular model airplanes
As someone into RC flight I'd love to see a break down of the components used for that build (batteries, ESCs, motors) as well as the specs (weight, flight time, max acceleration) and performance metrics (peak amps, temps, etc). As mentioned elsewhere in this thread, this kind of performance isn't that unusual in the ultra high-performance drone world. They are definitely building on the work of others but it's nice to see this kind of engineering get more mainstream exposure.
It's the props that are most interesting to me, as this is what drives a lot of the speed capability - they're some kind of extremely high pitch bi-blades, I think 5", but I haven't figured out which yet.
I'd be curious to see their peak current draw too, and how much trouble they have with heat running F80 Pros on 8S.
:D
Then proceeds to explain how a drone isn't designed for taking laps on a track.
F1 is by comparison the least accessible sporting path in the world. The drivers who make it are great but they represent a rounding error equivalent of a potential candidate pool.
A lot of these drivers debuted with karting, which while not being exactly cheap, is accessible to the most of the middle class.
> The split is approximately: $2m on karting, $500,000 per season in F4, another $1m on F3 and $3m plus on F2 – without contingencies. Not all of that may come from parents’ pockets, but seven million bucks is still a massive pile, particularly given there are no guarantees of an F1 seat at the end of it.
[1] https://www.racefans.net/2021/07/28/is-formula-2s-3-million-...
The F1 under Bernie Eccleston was criticized a lot by becoming "boring"; not only because a particular pilot's or their team's dominance, but the broadcast itself. When NASCAR had already live pilot tracking widgets and speed estimations on screen, F1 fans still looked at an outdated, pixelated racing car tyre animation, and basically zero "insights" beyond speed trap measurements.
Around 10 years ago, Liberty Media took over and now broadcasting and event marketing is "modernized". It became fun to watch races even when you are not cheering for the top pilots, simply because live data is available and better communicated, making the race more interesting to follow.
Netflix are going to use the same template to make shows about other forms of sport.
[1] https://www.youtube.com/watch?v=0nyYaL0dGAA , but there's plenty of those on YouTube