I'm curious how this compares to foam-frame designs. Being able to customize it is obviously a big advantage, as is the non-solid-infill of 3d-printed parts. I think for stiffness, 3d-printed frames don't work well for quadcopters compared to carbon fiber, but they sound like a nice alternative to foam for fixed-wing. I think the stiffness concern comes up in quads mainly when they do high-performance maneuvers that aren't a concern for the takeoff and landing this device does in that mode. (e.g. high accelerations/manevers of racing-style drones)
If anyone wants to try this: The parts he uses are all standard Chinese-made COTS you can buy on amazon and similar.
The ArduPilot firmware he uses is very flexible and robust, but setting it up is one of the worst UXs I've experienced. Commercial UASs almost universally use PX4 instead.
Yes I used single wall foaming PLA which is much less impact resistant and more brittle vs any foam, even cheap foamcore and especially EPP or EPO. This has definitely been an issue with crashing and rebuilding.
But my first and only other VTOL build was foamcore Readyboard and that took a 12 ft drop onto asphalt with only a slight compression in the fuselage. Never replaced it.
I would add dovetails or other clips for printed sections if I did another printed build.
Yes avionics and propulsion parts are COTS for speed, the Amprius pack is US manufactured but others are all made in China.
I'm starting to see some more Ardupilot used commercially too but yes the UX is janky and unintuitive.
Thanks, I did crash it again, and that's why it was half assembled on the wall at the end. But yeah foaming PLA is extremely soft, almost feels like tough, thin paper.
PLA is pretty bad compared to foam when it comes to planes, as it's very heavy and very brittle. Any semi-hard landing will break parts off, and heavy planes fly badly.
The big advantage is that you can just print the part again, which almost makes PLA worth it.
ABS would probably be better, as it's much more durable and lighter, but it's still much heavier than foam, and printing ABS isn't great.
I used single wall foaming PLA which has filament density of ~0.45 g/cm3 at 250 degree nozzle temp, about 64% lower than normal PLA. But it is even worse for impact resistance than normal PLA. Weight was the primary driver for this plane
Yep, I've used that as well, but, as you say, it's just very bad for impact resistance. I think the sweet spot between weight and durability might be ABS or something like ASA.
Yes and there is foaming ASA as well which is almost as light as foaming PLA. But I can't print ASA well on my A1 and the weight penalty was still enough that I didn't try, given the mission profile of flight time.
Printing ABS is much improved with the later hobbyist-level 3D printers (Qidi and Bambu etc). My Qidi Plus 4 prints ABS as easily / well as PLA. Just remember to do it in a well-ventilated space.
In case of commercial systems you are paying manufacturer to do the integration work and provide polished solution anyway, so setup UX not being user friendly isn't a major concern. In my opinion bigger factor for why most commercial UASs use PX4 instead of ArduPilot is licensing and PX4 maintainer friendliness towards commercial solutions. ArduPilot is GPLv3 and more geared towards community/hobbyists while PX4 is BSD. Commercial manufacturers don't want to disclose the source for their modified version of firmware with all the value added integrations they provide (or just deal with it even if their fork doesn't have anything interesting).
I have ~200 acres of land I fly a drone survey mission to map. Today, that piloting is done by dronelink and a DJI drone. The challenge is that it's about 3 hours of flight time, give or take, to cover the space, and a given battery is good for about 35 minutes of flight time.
I have 4 batteries, which means I basically need to be refilling my batteries as quickly as I consume them in order to keep flying continuously (unfortunately, even with a quad-charger, I cannot fully sustain this.)
I would LOVE to have a fixed wing drone that could fly over the area and snap close-to-nadir photos as it did so, but the complexity of building and programming a hand built drone seems so much higher than deploying an off the shelf DJI drone. Additionally, the land is steep, with a 1,000+ foot elevation difference, and rugged, and the neighboring land is airspace I cannot fly within, so I couldn't use it to perform my turns.
Author, or others, any thoughts on whether it's worth pursuing a fixed wing aircraft to perform this mapping mission? Or is the best bang for my buck to just buy enough batteries to fly the mission continuously on an off the shelf quadcopter?
Great question! I don't think there's any sub $5k VTOL you can just buy COTS that can get close to 3 hours of range. Also, nothing is as plug and play as DJI. If you're motivated enough to do a little DIY and learn how to use Ardupilot or PX4 (easier), you can buy a kit like the Heewing T2 VTOL and assemble yourself. But I don't think that will do more than two hours flight time, even with a similarly high energy density battery pack to what I used.
The FAA does not permit this, by default. You are allowed to remote pilot in command one drone at a time. I could apply for an FAA waiver (I have one to fly beyond VLOS). I've genuinely considered doing so, but I don't know how I could convince the FAA that I could safely monitor multiple drones simultaneously.
200 acres should not take 4 hours. It should be coverable within 20-25 minutes with 75-65 overlap, flying height 120m with something like Mavic 3 getting a GSD of 3.5cm/px. Look at optimizing your flight overlaps and height.
It depends a lot on the terrain type - highly repetitive terrain requires higher overlap. It does sound like they have some broken setting somewhere though.
It's heavily pine forested mountain areas. With a 65-75% overlap, the SFM algorithms struggle to produce sufficient details. Additionally, because of the verticality of the terrain and very tall pine trees there's a need to have multiple angles to generate a good orthophoto. So the grid is denser than other environments for a reason. I'm continuously updating my flight plan based on the results generated -- squeezing density up/down based on observed results.
The 4 hours is an overestimate, it's probably genuinely closer to 3 hours flight time.
The area was partially clearcut about a decade back. Some areas are due for brush management and some for commercial thinning. Additionally, because it is alpine and contains a stream used by fish for spawning, it is interesting to see the variations in snow load and water flow in the stream year over year.
So there's at least a reason to get out each winter (snow load), spring (melt/brush growth/flowers), and summer/fall (stream health/identify trees once brush loses leaves).
I also like seeing if there's trees in stands dying at an unusual rate, which might indicate pine beetle infestations or sickness that I'd need to take care of.
Also, it's a fun hobby and a cool dataset to flip through.
If you want and have time you can enter the world of FPV, building a drone yourself. You can change the frame, motors, ESCs, controllers etc. You have much more control compared to what you have with a DJI. It's very rewarding, too. But it takes time so it might not be the economic option.
It would be a project, but the HeeWing's T2 Cruza VTOL is extremely interesting as you could put in both a high capacity battery and a good quality camera.
Assuming 200 acres is 0.57x0.57 miles. If you fly at 15 mph you can do 15 lines in 30 mins. The lines would be 200 feet apart, and you'd move 22 feet per second. 15MP images would be 10 pixels per foot and motion blur with 240fps would be 1.1 inches.
0.57 x 0.57 miles is a good estimate. The lot is more or less square.
Yes. Maintaining 15mph is tough because of elevation changes. The land has a 1,200 foot tall ridge in it, so the drone must gain (and lose) 1,200 feet across that 0.57mi span. Which is steeper than the drone can do while maintaining a forward speed of 15 mph.
Additionally, SFM algorithms struggle with repetitive environments like pine forests, where there are not clear points or lines they can align. GPS data helps somewhat, but if the pictures are too sparse then it can get confused and start stitching images together incorrectly. Additionally, too few images lowers the fidelity of the output and causes strange gaps and black spots in clearings (where the clearing is visible in one pass but obscured in another.)
Finally, radio transmissions require two flights due to that ridge. I need to fly once on either side or I lose signal. There is no accessible point on the property I can maintain visual line of sight across the entire span, and "mountain" is a pretty good radio signal attenuator.
I'd LOVE to have some automation help in developing a flight plan, as right now it's manual, and I'm creating little segments myself, individually placing points and grids. Several tools claim to support automatic elevation, but none of them do so well when you have a 1000+ foot cliff or a narrow stream bed that has cut 20 feet into the earth in your environment.
I reviewed and purchased a smaller drone, from WHISPR. Through that, I looked at similar VTOL drones, which OP is demonstrating. In theory, they're cheap:
This one has a maximum climb rate of 3 degrees. Which means you'd need to plan a route that's continously climbing, and tacking back and forth to avoid going steeply. So you would probably combine flight planning using Mission Planning: https://ardupilot.org/planner/ and custom json tweaking following MavLink protocol: https://ardupilot.org/dev/docs/mavlink-routing-in-ardupilot....
The open source flight controllers and systems are https://ardupilot.org/copter/docs/common-autopilots.html#ope... a good place to start. There's a few orgs building the full experience, but you'll of course pay for fully integrated, but still open source, hardware+software.
The scenario you describe is exactly what a VTOL or normal play is for; quad copters simply do not cover enough ground.
Also consider the camera. DJI typically ship with really low quality cameras, but https://www.bhphotovideo.com/c/product/1785754-REG/sony_ilx_... would get you a much higher pixel density so if you're more interested in the imaging, you can flight higher, near the licensed ceiling, completing more ground faster.
I've also found AI models know how to calculate pixel densities, so it's pretty easy to mock out a flight plan even if you dont have the actual drone available.
While that Sony piece of kit probably can do some cool things, 2950$ without objective/lens, seems rather expensive. Also 8.6 oz / 243 g (Body Only) don't look that light to me.
I'd probably go for some Sony Experia 10III[¹] or upwards, for no more than 300$ at 170 g (unmodified), and strip the hell out of it. Does it need a display, all the parts of the case, and so on? Can can I use its GPS, and pipe it into the flight controller, instead of having to rely on something more powerful and heavy, power guzzling?
This actually has decent flight time! Could probably push close to 3 hours if using higher energy density (and cost) battery cells like the silicon anode ones I used.
How often do you need to do this mapping? if it’s every day it would definitely be worth it to look for another solution, if it’s every 6 months or so this seems tedious but doable.
The OP VTOL drone is aesthetically pleasing, I feel, but is over-engineered for a task such as mapping long distances.
For your needs, a slow-stick with a KFM wing would suffice - the smallest, lightest, simplest, most capable airframe one can build, and that is a slow-stick[0] with KFM wings[1].
VTOL is cool for those of us raised on a healthy sci-fi diet, but is entirely unnecessary for a slow-stick, which can be hand-launched directly into the air (vertically) and landed at a stall within a meter.
Plus, it is small, light, effective and extremely simple, which means its a lot easier to repair and maintain. Add some ardupilot magic and you've got a really capable platform for surveys and reconnaissance.
tl;dr Consider building a small fleet of slow-stick+KFM planes, attach your camera to it, and do your survey that way ..
Aesthetics follows function in this case. And agreed VTOL is not beneficial at hobbyist scale for most because of the ease and short distances required to hand/runway launch and land an RC plane
This is incredibly impressive. I'd love to hear more about what relevant skills & knowledge you started with, and how you worked out what you'd need to know to complete the project.
How much customisation did you need to make to Ardupilot? Is your drone's control unique, or somehow standardised?
Thanks! I'm using the standard Ardupilot control systems for hover, transition and cruise flight. On the firmware side, it's just some parameters and tuning that is customized.
What I started with
- Had built one VTOL before from foamboard, not 3D printed.
- Familiar with Ardupilot from that project and assembling a multicopter and COTS VTOL.
- So I had a little experience building a structurally sound airframe for VTOL loads, but 3D printing was a wrinkle.
- How I worked it out is a hard question. But it was being focused with design, flight testing and troubleshooting. LLMs, Youtube, forums, etc for help when needed.
- Building in public helped paradoxically. It actually saves me time to build in public because of the motivation boost that helps me move faster and to share progress sooner. Even though there's a higher lift to document and share.
Ardupilot is very, very mature software. A lot of the drone video coming out of Ukraine, the HUD overlay is likely from Ardupilot. If you can think of it, Arduipilot supports it. Airplanes, helicopter, VTOL, speedboat, even sailboats.
As someone who works with ArduPilot professionally, I have very mixed feelings about it. It’s mature, definitely. It supports all kinds of vehicles, like you say. It’s beautifully modular and supports a crapload of flight control and sensor hardware. And there are definitely pieces of it that are aggravating and exceptionally janky.
The HUD overlay you’re referring to is technically Mission Planner (GCS software), not ArduPilot (flight control software). Mission Planner and ArduPilot both talk Mavlink, and they’re both developed by the same community. MP is flexible. You can set it up to do almost anything you’d ever want. It’s also terrible and exceptionally janky… but extremely powerful. And they’re both free.
I think the problem with both of them is that they’re good enough that there isn’t likely to be a huge critical mass developing a better alternative. On the GCS side there is also QGroundControl and APM Planner 2 (which was a fork or reimplementation of Mission Planner). Both of them have their own upsides and downsides, but neither one of them is as mature or as powerful/flexible as Mission Planner. PX4 on the flight controller side is popular commercially because it’s BSD-licensed instead of GPL, but the net result is that it has nowhere near as many features as ArduPilot because companies build proprietary features and don’t push them back upstream.
This stuff is definitely in the worse-is-better domain. ArduPilot is free, ArduPilot is amazing and ArduPilot sucks. :)
Anyway, off to bed. I’m in a long test campaign right now and we’ve got to be up at 0430 to fly the ArduPilot-based aircraft again before the weather goes sideways.
Yes, one of the challenges though is airframe structural integrity when building it yourself. Need to have good enough understanding of multicopter and fixed wing plane design
I’m curious about the control surfaces. Since you have the four quadcopter motors, you could potentially just induce all three of yaw pitch and roll by powering those up.
I wonder if the reduction in weight from the now unneeded servos would pay for the extra battery drain.
Great question, but spinning lifting motors in cruise draws more power than it's worth for efficiency.
There's ways to use differential thrust for multiple cruise motors to roll if not positioned at spanwise Cg (e.g. wing mounted like in a commercial jet), but again, usually not efficient. Servos are a small mass fraction relative to other components.
Very cool. Very impressive. I hope it inspires other to build things they're passionate about. "You can just do things. And learn things." No need to wait for permission, or classwork (much less a degree), or a guide/teacher.
Thanks and yes I'm glad you say that. Would love to see it. It's hard to convey how much faster and better your work will be if you are actually passionate about it.
As someone who has always been curious about building one but haven’t dove into it, I’d love detailed plans and a beginner-oriented tutorial. I’d be happy to donate/patreon to an effort.
This is the first I've heard of foaming PLA, definitely have to check that out.
Did you do the whole airplane with a small printer like the A1 / A1 mini? I would love to print airfoils but I'm struggling to imagine a way to link individual prints together in a way that preserves stiffness. My 100cm wing would need 10x (10x10cm) printed parts somehow attached to one another.
Until I figure this out it's foamboard building for the type of airplanes I want to build (glider)
A1, 256 x 256 mm lets me print these wings in four sections (including a double walled thin section for the boom mounts). With a high AR glider you'd probably need a few more. Carbon fiber spars + CA glue will do the trick for attaching them
Yeah I printed one of the Titan Dynamics airframes (before they closed up at access for the hobby market) on a Prusa MK3S using foaming PLA and CF spars. The fuselage came out in three prints, the wings as one print each, and then the two wing tip extensions got combined into one print.
It would definitely have been tight on an A1 mini but the full size A1 would work great for sure.
I actually managed to batch print the wing sections which you can see in the video in the rebuild chapter. That works really well for wing sections because each takes up a minimal bed area
Definitely! I would have done the same but it was my first time working with the Foaming PLA filament and didn’t want to waste too much of it if the print failed.
Thanks, I've not! But given the battery mass fraction is 53% for this build, rule of thumb is about 30% of a VTOL's weight for battery if you want to include payload. So I could just use a smaller 6S battery.
Or I could probably add another ~0.5 lbs or a little more without issue. The lifting motors hover at 45% ish throttle so there is some headroom for more payload without reducing battery mass fraction.
I loved it! But I loved even more the idea of when a person has no prior experience or knowledge in something yet achieves great results in a short period of time. That build mindset is fascinating. The only issue is when applying for normal jobs, unless the interviewer can see through a person's passion, they usually assume you don’t have what it takes or won’t "raise the bar" for whatever they are after. That is, unless you lie or throw in all sorts of buzzwords to create a halo effect.
I have a question, though: Any info about the flight stack? Was it Pixhawk/Ardu, iNav, or something else?
Ardupilot is really cool, and a lot of people use Mission Planner with it. What do people think of Mission Planner, and are there any other software options people enjoy using?
Nice work! This is very impressive and you’ve shown real resilience and perseverance to work through the challenges you encountered while building this.
> That battery seems expensive (most expensive part?)
At first, I was like "Nah, just looks like my RC car battery", but then I Googled it...and it's a whopping $1,305 [0].
I looked closer at the specs and they're insane for a battery as small as it is. They have double the energy density as measured by both volume and mass compared to my RC car batteries [1].
440.64 Wh, 21600 mAh/20.4V, 1250W/60A constant discharge, 170A burst for 10s, total pack weight 1332g-1335g depending on connectors which gets you to 360 Wh/kg cell density or 330 Wh/kg pack density.
Impressive, but not that far off from much cheaper mass-produced cells.
He has separate motors for vertical and horizontal flights, which simplifies the design, but creates a rather bad inefficiency, the vertical motors create lots of drag during the horizontal flight.
Maybe it's not a big deal, I'm not sure. Making motors rotate would add weight for sure, thus reducing the range.
With this config, the cruise motors and prop are optimally sized for cruise - which gives non trivial gains to both eta for propulsive motor efficiency and prop efficiency.
Vs a tiltrotor/wing/body in which the cruise motor has to do double duty as lifting motors. Given it takes anywhere from ~4-7x more power to hover (depending on disc loading) than to cruise, you can see how the motors are not in an optimal throttle/rpm band in this case. Archer's CTO Munoz has actually said this publicly.
Very similar design already used by Wing. I'm guessing they did a fair bit of analysis and modeling of the cost, range, complexity, safety, etc. etc. tradeoffs before settling on what they're using currently.
Two impressive things about Wing's design are
1) load paths are designed to break the airframe in controlled ways
2) the 4 blade props have alternating shorter and longer blades for quieter aeroacoustics
Adam Savage did a video tour of their factory recently, worth a watch
Adding a tiltrotor mechanism is surely not worth the added complexity and weight, in this case. You're right though on the added weight and drag from having separate motors and props.
Tilt-rotor on all 4 motors with an extra twist: the wing shape adds to the lift in vertical mode, so you can use smaller motors, so they're more efficient even in horizontal mode.
Yes it really depends on the transmitter protocol. I was using SBUS, simple, cheap, familiar from my previous project. But limited to < 1 mile or so I believe, but ELRS and other protocols allow tens of miles or more.
Impressive! Can you share the design to the Ukrainians? [be careful, so it won't leak anywhere else]. 3 hours loitering can mean a lot on a battlefield.
Not tested! But minimum I would say 70 mph. If designing for speed, I'd be looking at a cruise motor and prop that maximizes the thrust output rather than cruise efficiency
Could be right! Really depends because the quad motors on freestyle say have massive throttle upside they can tap but it increases power draw and reduces flight time by multiples.
Yea def! You could also get into questions of cruise speed with level altitude vs in a dive etc, and you burn through batteries fast when max-performing quads!
Exactly! Didn't even mention cruise vs dive. But the latest WR for a quad doing 347mph I think, is just insane, even if drawing multi hundred amps for a few seconds only.
Oh wow, that’s even higher than this one I saw from last year @ 298 MPH for a quadcopter, but I’m not sure if this is the same criteria for the WR or build limitations etc?
Funny nitpick, this definition applies to most drones, because most drones sold are x-copters and do not have wings, they always take-off and land vertically.
Yeah it's a strange term because it probably originated relative to fixed wing planes. Ie a VTOL plane. But now multicopters are the predominant species so VTOL can sound redundant to most drone builders today.
Ukraine has dozens of drone producers, many are capable of producing 10k+ per month. This one is quite expensive (2000$) and more is for home-grown terrorist (like 3d-printed firearms)
Simplified they just get thousands of commands from which position to which position to move a toolhead, at what speed and how much plastic to extrude.
They would have to re-create the 3d model from those commands and compare them to banned ones? But then the firmware is often open source or at least can be flashed. And also how would the printer know if it's a toy gun or a real gun? Or just a part for a gun.
Paper printers / copiers refuse to copy money because the government controls the source, the paper money. They add small dots on them.
The government can't add those dots to self made 3d models.
>>Andy Greenberg has been reporting on ghost guns for more than a decade. He first used a 3D printer to assemble a gun in 2015, and he says that today’s process is not only faster but cheaper. We talk to Andy about how he legally printed the same gun Luigi Mangione allegedly used in the alleged killing of the United Healthcare CEO last year, and whether US law is keeping up with the technology of 3D-printed guns.
The "explody bit" that needs to survive intensive pressures, which boils down to a carefully machined barrel (rifling is optional) and a firing pin.
The "comfort features" that make it easy to hold that barrel in a human hand, that direct the recoil to parts of the human body that can better handle it, that hold the ammo in a way that automatically feeds the next bullet (spring magazine) and that makes it easy to press the firing pin against the munition (the trigger).
Yes, this is a thing. Supposedly such a thing was used in a recent attack. 3D printers are not able to determine intended operation of their printed output from their print job inputs, whereas counterfeit money has no legitimate purpose, so anything that looks “close enough” for an automated system to detect is probably close enough to fool a human if passed to them as if it were the real thing. I think preventing counterfeit money is a broad enough social good/benefit that a broad cross section of industry, government, and private individuals and groups have decided to work together to make it harder to produce. I don’t see who is harmed by this, besides people trying to print their own currencies for their own purposes, some of whom aren’t trying to make counterfeit copies of actually-existing currencies, but I think they are already going to have to worry about the counterfeit problems themselves if their currency is used outside of controlled conditions by trusted parties.
If anyone wants to try this: The parts he uses are all standard Chinese-made COTS you can buy on amazon and similar.
The ArduPilot firmware he uses is very flexible and robust, but setting it up is one of the worst UXs I've experienced. Commercial UASs almost universally use PX4 instead.
But my first and only other VTOL build was foamcore Readyboard and that took a 12 ft drop onto asphalt with only a slight compression in the fuselage. Never replaced it.
I would add dovetails or other clips for printed sections if I did another printed build.
Yes avionics and propulsion parts are COTS for speed, the Amprius pack is US manufactured but others are all made in China.
I'm starting to see some more Ardupilot used commercially too but yes the UX is janky and unintuitive.
The big advantage is that you can just print the part again, which almost makes PLA worth it.
ABS would probably be better, as it's much more durable and lighter, but it's still much heavier than foam, and printing ABS isn't great.
I have 4 batteries, which means I basically need to be refilling my batteries as quickly as I consume them in order to keep flying continuously (unfortunately, even with a quad-charger, I cannot fully sustain this.)
I would LOVE to have a fixed wing drone that could fly over the area and snap close-to-nadir photos as it did so, but the complexity of building and programming a hand built drone seems so much higher than deploying an off the shelf DJI drone. Additionally, the land is steep, with a 1,000+ foot elevation difference, and rugged, and the neighboring land is airspace I cannot fly within, so I couldn't use it to perform my turns.
Author, or others, any thoughts on whether it's worth pursuing a fixed wing aircraft to perform this mapping mission? Or is the best bang for my buck to just buy enough batteries to fly the mission continuously on an off the shelf quadcopter?
The 4 hours is an overestimate, it's probably genuinely closer to 3 hours flight time.
So there's at least a reason to get out each winter (snow load), spring (melt/brush growth/flowers), and summer/fall (stream health/identify trees once brush loses leaves).
I also like seeing if there's trees in stands dying at an unusual rate, which might indicate pine beetle infestations or sickness that I'd need to take care of.
Also, it's a fun hobby and a cool dataset to flip through.
Never used one, but either this design or a tri-rotor v-22 style tilt motor are the two designs I've found intriguing. Worth checking out at perhaps?
Anything wrong with that?
Yes. Maintaining 15mph is tough because of elevation changes. The land has a 1,200 foot tall ridge in it, so the drone must gain (and lose) 1,200 feet across that 0.57mi span. Which is steeper than the drone can do while maintaining a forward speed of 15 mph.
Additionally, SFM algorithms struggle with repetitive environments like pine forests, where there are not clear points or lines they can align. GPS data helps somewhat, but if the pictures are too sparse then it can get confused and start stitching images together incorrectly. Additionally, too few images lowers the fidelity of the output and causes strange gaps and black spots in clearings (where the clearing is visible in one pass but obscured in another.)
Finally, radio transmissions require two flights due to that ridge. I need to fly once on either side or I lose signal. There is no accessible point on the property I can maintain visual line of sight across the entire span, and "mountain" is a pretty good radio signal attenuator.
I'd LOVE to have some automation help in developing a flight plan, as right now it's manual, and I'm creating little segments myself, individually placing points and grids. Several tools claim to support automatic elevation, but none of them do so well when you have a 1000+ foot cliff or a narrow stream bed that has cut 20 feet into the earth in your environment.
https://www.uavmodel.com/products/makeflyeasy-hero-2180mm-ua...
This one has a maximum climb rate of 3 degrees. Which means you'd need to plan a route that's continously climbing, and tacking back and forth to avoid going steeply. So you would probably combine flight planning using Mission Planning: https://ardupilot.org/planner/ and custom json tweaking following MavLink protocol: https://ardupilot.org/dev/docs/mavlink-routing-in-ardupilot....
The open source flight controllers and systems are https://ardupilot.org/copter/docs/common-autopilots.html#ope... a good place to start. There's a few orgs building the full experience, but you'll of course pay for fully integrated, but still open source, hardware+software.
The scenario you describe is exactly what a VTOL or normal play is for; quad copters simply do not cover enough ground.
Also consider the camera. DJI typically ship with really low quality cameras, but https://www.bhphotovideo.com/c/product/1785754-REG/sony_ilx_... would get you a much higher pixel density so if you're more interested in the imaging, you can flight higher, near the licensed ceiling, completing more ground faster.
I've also found AI models know how to calculate pixel densities, so it's pretty easy to mock out a flight plan even if you dont have the actual drone available.
I'd probably go for some Sony Experia 10III[¹] or upwards, for no more than 300$ at 170 g (unmodified), and strip the hell out of it. Does it need a display, all the parts of the case, and so on? Can can I use its GPS, and pipe it into the flight controller, instead of having to rely on something more powerful and heavy, power guzzling?
[¹] because https://xdaforums.com/f/sony-xperia-10-iii.12225/
cf. https://en.wikipedia.org/wiki/Ingenuity_(helicopter)
I dont know much about the weights involved, but hardware surgery was not interesting to me.
Keepbin mind tge guys starting at DJI and youre proposing stripping dowm a camera
There really is a full open source hardware/software path, which is heartening.
For your needs, a slow-stick with a KFM wing would suffice - the smallest, lightest, simplest, most capable airframe one can build, and that is a slow-stick[0] with KFM wings[1].
VTOL is cool for those of us raised on a healthy sci-fi diet, but is entirely unnecessary for a slow-stick, which can be hand-launched directly into the air (vertically) and landed at a stall within a meter.
Plus, it is small, light, effective and extremely simple, which means its a lot easier to repair and maintain. Add some ardupilot magic and you've got a really capable platform for surveys and reconnaissance.
tl;dr Consider building a small fleet of slow-stick+KFM planes, attach your camera to it, and do your survey that way ..
[0] - https://www.rcgroups.com/forums/showthread.php?1395335-Begin...
[1] - https://www.flitetest.com/articles/kfm-wings-a-basic-explana...
Not sure I agree with you here. The very same functionality can be attained with far less plastic.
How much customisation did you need to make to Ardupilot? Is your drone's control unique, or somehow standardised?
What I started with - Had built one VTOL before from foamboard, not 3D printed. - Familiar with Ardupilot from that project and assembling a multicopter and COTS VTOL. - So I had a little experience building a structurally sound airframe for VTOL loads, but 3D printing was a wrinkle. - How I worked it out is a hard question. But it was being focused with design, flight testing and troubleshooting. LLMs, Youtube, forums, etc for help when needed. - Building in public helped paradoxically. It actually saves me time to build in public because of the motivation boost that helps me move faster and to share progress sooner. Even though there's a higher lift to document and share.
The HUD overlay you’re referring to is technically Mission Planner (GCS software), not ArduPilot (flight control software). Mission Planner and ArduPilot both talk Mavlink, and they’re both developed by the same community. MP is flexible. You can set it up to do almost anything you’d ever want. It’s also terrible and exceptionally janky… but extremely powerful. And they’re both free.
I think the problem with both of them is that they’re good enough that there isn’t likely to be a huge critical mass developing a better alternative. On the GCS side there is also QGroundControl and APM Planner 2 (which was a fork or reimplementation of Mission Planner). Both of them have their own upsides and downsides, but neither one of them is as mature or as powerful/flexible as Mission Planner. PX4 on the flight controller side is popular commercially because it’s BSD-licensed instead of GPL, but the net result is that it has nowhere near as many features as ArduPilot because companies build proprietary features and don’t push them back upstream.
This stuff is definitely in the worse-is-better domain. ArduPilot is free, ArduPilot is amazing and ArduPilot sucks. :)
Anyway, off to bed. I’m in a long test campaign right now and we’ve got to be up at 0430 to fly the ArduPilot-based aircraft again before the weather goes sideways.
I wonder if the reduction in weight from the now unneeded servos would pay for the extra battery drain.
There's ways to use differential thrust for multiple cruise motors to roll if not positioned at spanwise Cg (e.g. wing mounted like in a commercial jet), but again, usually not efficient. Servos are a small mass fraction relative to other components.
Net effect is yaw counter clockwise, but zero net change in lift.
Incredibly humbling! “””
The imagination to reality loop is most rapidly well-tuned for categories that exist.
Did you do the whole airplane with a small printer like the A1 / A1 mini? I would love to print airfoils but I'm struggling to imagine a way to link individual prints together in a way that preserves stiffness. My 100cm wing would need 10x (10x10cm) printed parts somehow attached to one another.
Until I figure this out it's foamboard building for the type of airplanes I want to build (glider)
It would definitely have been tight on an A1 mini but the full size A1 would work great for sure.
Or I could probably add another ~0.5 lbs or a little more without issue. The lifting motors hover at 45% ish throttle so there is some headroom for more payload without reducing battery mass fraction.
I have a question, though: Any info about the flight stack? Was it Pixhawk/Ardu, iNav, or something else?
https://www.youtube.com/watch?v=Dd2N_lyO_SQ
https://ardupilot.org/planner/ - the website seems down right now though.
And if you're looking for other ground controls, you could try out QGroundControl or Mavproxy (terminal based).
At first, I was like "Nah, just looks like my RC car battery", but then I Googled it...and it's a whopping $1,305 [0].
I looked closer at the specs and they're insane for a battery as small as it is. They have double the energy density as measured by both volume and mass compared to my RC car batteries [1].
[0] https://www.upgradeenergytech.com/product-page/gold-v1-6s2p-...
[1] https://www.amazon.com/Zeee-Connector-Vehicles-Helicopter-Ai...
Impressive, but not that far off from much cheaper mass-produced cells.
He has separate motors for vertical and horizontal flights, which simplifies the design, but creates a rather bad inefficiency, the vertical motors create lots of drag during the horizontal flight.
Maybe it's not a big deal, I'm not sure. Making motors rotate would add weight for sure, thus reducing the range.
With this config, the cruise motors and prop are optimally sized for cruise - which gives non trivial gains to both eta for propulsive motor efficiency and prop efficiency.
Vs a tiltrotor/wing/body in which the cruise motor has to do double duty as lifting motors. Given it takes anywhere from ~4-7x more power to hover (depending on disc loading) than to cruise, you can see how the motors are not in an optimal throttle/rpm band in this case. Archer's CTO Munoz has actually said this publicly.
https://en.wikipedia.org/wiki/Wing_Aviation#/media/File:Wing...
Adam Savage did a video tour of their factory recently, worth a watch
Thanks! https://www.youtube.com/watch?v=_BXm6dTHvY0
I'm a big Waymo and Wing fan, but hadn't seen that.
Tilt-rotor on all 4 motors with an extra twist: the wing shape adds to the lift in vertical mode, so you can use smaller motors, so they're more efficient even in horizontal mode.
> the wing shape adds to the lift in vertical mode
Wings require airspeed to work, which there presumably aren't a lot of in vertical mode.
This is also the most the penalty will ever be as electric motors continue increasing in specific power, and quite rapidly this century so far.
Either way, nice work!
Would love to try my hand at reproducing the work you did!
hope I can build mine 3d printer lab first
edit: at least 70 mph
https://www.youtube.com/watch?v=wThmg8Ezm9w
This should be "a 3D printed winged VTOL drone"
is this even a thing lol???
I thought manufacturer already have a safety in place like printer that refuse to print money for that
Simplified they just get thousands of commands from which position to which position to move a toolhead, at what speed and how much plastic to extrude.
They would have to re-create the 3d model from those commands and compare them to banned ones? But then the firmware is often open source or at least can be flashed. And also how would the printer know if it's a toy gun or a real gun? Or just a part for a gun.
Paper printers / copiers refuse to copy money because the government controls the source, the paper money. They add small dots on them.
The government can't add those dots to self made 3d models.
Here is a link: https://www.wired.com/story/uncanny-valley-3d-printed-untrac...
The "explody bit" that needs to survive intensive pressures, which boils down to a carefully machined barrel (rifling is optional) and a firing pin.
The "comfort features" that make it easy to hold that barrel in a human hand, that direct the recoil to parts of the human body that can better handle it, that hold the ammo in a way that automatically feeds the next bullet (spring magazine) and that makes it easy to press the firing pin against the munition (the trigger).
The latter can easily be 3D printed.