One option would be a Navio2 [1] flight controller which runs the PX4 or Ardupilot flight stack on a realtime linux kernel on an rpi. You'd then need to put that on an airframe (a F450 or one of it's many clones is a solid choice if you intend to use this thing outdoors).

You'll also need all the ancillaries (receiver, batteries, charger, rc transmitter) although you'll need those anyway if you plan to fly outdoors in a safe manner.

Alternatively, if you want a ready to go solution which is good for indoor and limited outdoor use, the Parrot ARDrone [2] runs linux, has two cameras and has aftermarket firmware available from both Ardupilot and Paparazzi projects.

Be aware that if you intend to use your drone for 'work' or commerce you should check your local air regulator (FAA/CAA/CASA etc) to see whether you require any permissions or qualifications to legally operate.

1: https://emlid.com/navio/

2: https://www.parrot.com/uk/drones/parrot-ardrone-20-gps-editi...

I have the crazyflie nano, comes in at $180. Camera would be extra, but not much more. You can get 1k in range with the antenna upgrade. There are also many other neat upgrades like inductive charging and indoor positioning. Buy spare motors and clips, or a 3d printer. Having some guards while you work on flight algos will be helpful. Try out some RL

https://www.bitcraze.io/crazyflie-2/

This is a good one, the kind i was looking for. However, I couldn't find the camera option. Are you sure the integration is possible? Do you have any links?

Best I have found...

https://forum.bitcraze.io/viewtopic.php?f=6&p=8295

The Qualcomm Snapdragon Flight and the Intel Aero systems are going to come in at like $500-$1000. If you want to process camera frames, your price target is a bit untenable for now. You might try to use any old flight controller and a pi with a camera, or a cheap wifi drone like the qr-w100s and doing computer vision over wifi (here's a repo we started for this: https://github.com/SeneCameras/qr-w100s)

If you're considering crazyflie, Flybrix is a programmable octocopter at $189. We are developing a camera module with onboard processing for optical flow, to release later this year. we are taking extraordinary pains to get a programmable drone with programmable computer vision at a reasonable price point.

Nvidia Jetson TX1/TX2 is available now for right around $400 with 256 CUDA cores 4-8GB of ram and 16-32GB of disk.

There are drones available and in planning based around the TX1 (and presumably the TX2). Note that because of the power requirements, the drone size is larger.

I would definitely love to get one based on a TX2. It's the only chip that has enough computing power to do real time image processing, run SLAM and a small CNN at the same time.

Would deffo pay 500-1000 for it.

I'm not an expert but I have quite a bit of experience in the RC world.

I haven't seen anything that fits those requirements in that price range.

Just getting a PixHawk flight controller so you can program the drone would cost that much. Same for a good radio transmitter to actually fly the drone.

I think DIY is out of the question at that price point.

"PixHawk flight controller"

That's a nice controller, but heavy and expensive, 37 grams and $75 according to amazon. My sub-250g build used a Naze32 which weighs about 4 grams without connectors and was $35. I flashed it with cleanflight from github

https://github.com/cleanflight/cleanflight

Presumably OP wanting to modify stuff will want FOSS like cleanflight.

This is of course ancient obsolete history because things move fast in multicopter world and that was like 6 months ago. Not even sarcasm. This advice was reasonably contemporary last fall, but this spring I donno.

OP will rapidly get into a scalability problem where I assume OP wants to do something interesting or complicated and aerospace is highly optimized so OP is likely to have much more success driving a flight controller as if it servos using a ras-pi or something, than trying to compile cleanflight to add Alexa to the cleanflight itself or whatever it is OP is trying to do.

Electronics have improved quite a bit. Back in the 80s we'd buy a $200 mosfet motor controller with a half pound heatsink and connect it to a $20 motor in your RC dune buggy (running off old fashioned nicads LOL) but now a days you attach a $11 no-heatsink controller to a $40 brushless motor and call it good. Of course you don't need a cooling fan if you have hundreds of watts of fan blowing on it from half an inch away...

Something to think about ops desire for a $100 multicopter is control electronics and surprisingly chassis and props and serial receivers are cheap, but I dropped $40 on EACH motor, by far the most expensive part of the build. Of course at 220 grams flying weight my build has the vertical acceleration performance of a large model rocket, which can be slightly terrifying when learning to fly, something lower performance but still flyable might be cheaper. I just thought it interesting that most of my build's money is in the motors, like about half of it.

Do you have any videos of the acceleration? That sounds awesome.

No, but while learning to fly I'd describe it more as terrifying. OK big wind gust or whatever we're getting too low, too low, too low, floor it, ... and a second later we're 50 feet up moving very fast upward and that's the new problem.

Pilot induced oscillation, never fun.

I can hover around 1/3 throttle so that would imply 1 G of thrust so you can guess that flooring it would be 2 G vertical acceleration. So 64 ft/s squared so one second after I floor it I'm going 40 mph straight up.. It feels like throwing a $300 volleyball as hard as you can straight up.

To stay underweight I use small 500 mAH lithium poly batteries which lead to both high acceleration and short flight times, like 3-4 minutes tops. I think I have a 250 somewhere which is really small and light. I could probably physically lift 2000 mAH batteries or larger, but would then be over the FAA 250 gram limit so I'd have to register as a drone pilot, which I'm OK with, other than I haven't done it yet. Great, another government list to find myself on, just what I needed. Once you're on enough lists, one more won't hurt...

That's enlightening. Do you know anything with a higher price point? Like around $500 ?

As a DIY drone builder for 5 years, here are my two pennies.

Arduino lets you program it. Apm and multiwii are essentially arduino. Video is a bit tricky, but you can always transmit it back to your laptop and process however you like

5 years??! That's amazing. Do you have a blog about your projects? That would be an interesting read.

Previous responses:

2016 https://news.ycombinator.com/item?id=10872233

2015 https://news.ycombinator.com/item?id=8989411

Tangential, for whoever also wonders this:

10872233 - 8989411 = 1882822

This is somewhat offtopic to the OP's question but I've been wondering about something mildly related for some time.

What sort of price range would I be looking at for a homemade long-range drone (I assume a plane) that could fly fairly high for a while and carry 10kg+ of weight?

I ask this as someone completely ignorant of aviation; I'm not sure what sane values of "fairly high" and "fly for a while" actually are.

For "fairly high" I'd say "can it be high up enough to be hard to see and hear from the ground?" (Binoculars are fine; camera zoom lenses win everything and trying to fight those would be stupid.)

For "fly for a while" I'm curious if I'd be able to fit enough gas (I don't expect batteries to work for this) to run for 20 minutes, or even an hour or maybe more.

I'd be interested in what a cost gradient for 10kg, 20kg, etc would look like.

This is just a "how much would it cost to carry things around" back-of-my-mind question I've been mentally noodling for a while.

I don't honestly need super-specifics; back-of-the-envelope ballpark figures would be great. (I expect Amazon will nail drone delivery in practice; not trying to plan business ideas or anything.)

--

Also - what's out there that can fly for an hour+, and which is reasonably "positionable", as in something I can get to hover in place for short periods? (ie, a plane needs to fly straight to stay in the sky, so that doesn't work)

I completely disagree with the other reply to your post.

Keep in mind you said drone and not quadcopter. You could very easily buy a ready to run flying wing [1] with a 30 minute flight time out of the box for just $90. A flying wing is more more suitable for gas or jet A powered motors as the natural lift of the airframe shape keeps it up in the air and balanced. This design scales up linearly (see the B-2 bomber.)

I've hypothesized that a dirigible filled with helium or hydrogen could potentially stay up for days or weeks if you had a full carbon composite airframe with a well sealed mylar skin. Lift capacity wouldn't be as great but it's still a fun thought exercise.

[1] https://hobbyking.com/en_us/phantom-fpv-flying-wing-epo-airp...

Basically all civilian fixed wing aircraft have an order of magnitude less thrust than what they'd need for flying straight up. Wings are an awesome piece of tech!

Google heavy lift drones. The specs you are looking for are payload and flight time. They are going to be inversely related.

Expect to pay more than $10,000 for an off the shelf solution. 10kg is a lot of weight for a commercial drone. Are you sure you need that much?

I see.

I'm mostly finding drones with multiple rotors and batteries thus far, not much that's like a more traditional "is a plane, has an engine" type of approach, which I suspect will net me a fair bit of flight time.

This is really just a thought experiment - I saw the thread and decided to throw the question out there as something I've been mildly curious about for some time. "Could I load up a bunch of stuff into something that looks like a (big) model plane, head down to the local park, hit 'go' and have it land near a friend's house in a couple hours?"

I'll admit I was kind of expecting $10k, although that is duly disappointing to learn nonetheless.

I think "long range" probably will violate many countries drone laws. I'd investigate the legality before proceeding with a build project.

The FAA doesn't care about range but is pretty excited about altitude (keeping it low), being away from airports, and being under direct human observation/control or if doing first person video with goggles having some kind of backup.

The FCC is who gets annoyed when long distance folks do the "CB linear" attitude and basically ignore all power regulations to get range.

Most of the civilians doing research and "production" flights currently are more likely to be violating various DEA laws with somewhat more draconian penalties than the FAA or FCC would every enforce anyway. 10 Kg of coke is going to be problematic to talk your way out of.

My experience with traditional RC planes is anything liquid fueled is audible for at least half a mile unless you go to some effort, but electric assist gliders, although huge and highly visible, are nearly inaudible maybe 10 feet away unless you're doing something dumb (running the prop tips supersonic to get more thrust, etc). A glider would also have much better range. Lets just say you should be wearing hearing protection when tuning a 0.40 sized RC airplane engine without a muffler, and probably while a muffler is installed too. Its a lot louder than you'd expect based on experience with yard work equipment (other than a chainsaw... A chainsaw is a pretty good comparison...)

My multicopter sounds like a weedwhacker in operation, and I can easily hear the 5volt beeper go off 30 feet away when the battery voltage drops too low. I donno maybe 50 feet away you wouldn't notice it? This is a 210mm sub 250g build, I'm sure a monster sized aircraft has monster sized noise to match.

It's hard to answer this question without knowing more precise values for "good range", "reasonable flying time", and even "ability to be programmed". At this price point, I think your only option is to buy an off-the-shelf quadcopter and modify it yourself. The $30 Hubsan H107[0] can be reflashed with an open-source firmware[1], although you won't be able to process video from the camera using the onboard MCU.

[0] https://www.banggood.com/Hubsan-X4-H107C-2_4G-4CH-RC-Quadcop...

[1] https://github.com/goebish/bradwii-X4

If you want low-level access to the hardware for reprogramming the stability routines, then any hobby racing drone running the CleanFlight or BetaFlight (https://github.com/betaflight/betaflight) software would be good... They can be built for under $100. However the camera won't be accessible by the flight controller.

If you only need high-level access to a flight API - e.g. sending commands like "take off" and "forward 1m", then take a look at GoBot: https://gobot.io/documentation/platforms/ - they have a few quadcopters on their supported platforms list.

The DJI Mavic Pro runs Linux. Some are working on reversing their firmware.

https://forums.hak5.org/index.php?/topic/39735-reversing-mav...

FWIW, don't know if this is mentioned in earlier threads and I think the price is higher but I think it fits into this thread anyway (Norwegian, so you might want to run it through translation, but has pictures, videos and parts lists):

https://nrkbeta.no/2011/10/13/slik-bygger-du-ditt-eget-quadk...

I recently came across the following custom Eachine E010S build [0] based on a Beecore F3, which can be programmed with e.g. Betaflight, although the video processing would have to happen on the receiver side...

[0] https://youtu.be/3ED_jCKlhP8

Tangent to the OP's question, any up to date database of the most relevant drones in commerce? I am advising a firm who is developing a testing bench for the diagnosis of drones and some numbers or market parameters would be welcome. Thanks.

Ardupilot on MEGA2560 with a Pi/Odroid for any video processing would be fine. You can get away with a 3500mAh battery on a quad w/ 1045 props & 1000KV motors for 15-20mins.

PM me for another great combination/consultation.

Ardupilot team has stopped firmware development for APM1/APM2 boards in 2015. I wouldn't advise buying it now with intent to work on stabilization firmware.

Yeah but stabilisation doesn't change over time, for the most part it still remains PID based. 8-bit MCUs are so under-utilised & 32-bit processors are simply an overkill & barrier to people getting started.

The attitude stabilization- yes, I would agree.

But the development mostly seems to concentrate around higher level, position and navigation controllers and there is a lot of complex floating point math going on there.

Interesting. How do I pm you?

ap46in gmail

I know this isn't quite the kind of answer you're looking for, but for those who are willing to put even more work in, $150 should be able to get you reasonably far with hobbiest electronics and a 3D printer (that price is obviously not including the cost of purchasing a printer).

Take for example, this open source chassis [1] that is based on the DJI Mavic (retail $1,000). Plenty of fun for those with the time and desire to take this even further.

[1] http://www.thingiverse.com/thing:2004357

I think you underestimate the price of electronics...

I doubt you'll succeed with $150... but for programability, range, flying time & decent payload... I'd consider a Quad VTOL aerofoil....

Such as: http://px4.io/portfolio/deltaquad-vtol/

Not sure about reasonably priced UAVs, but the AscTec Hummingbird is high on the state of the art ranking:

http://www.asctec.de/en/uav-uas-drones-rpas-roav/asctec-humm...

There is nothing for that money but if you want to DIY a programmable UAV, start here:

http://ardupilot.org/

It's hard to recommend anything without knowing the specifics of what you want to use it for

Phantom 3 standard is about 330 bucks. You can use the DJI SDK to program it.

At 150 its going to be quite hard to get all the features you are looking for. Even with DIY route you will need Motors, Controller, Camera, Transmitter etc. Easily 300+ bucks.

Cool but where can you get Phantom 3 standard for 330 USD? I'm seeing $499 on their official website.

Pretty much any ARM based flight controller board would be programmable...

A bit more expensive than what you are looking for, but I got the DIY (not built) 3D Robotics Y6 for $600 which runs the PX4/Ardupilot stack. I have been able to use it with ROS.

About "reasonable flying time" - are there any options available to use solar panels? Most importantly are solar panels good enough to power the unit?

A quick sanity check is that there aren't many solar-powered flying creatures.

A 15W solar panel is about three square feet, and probably weighs several pounds. A 15W quadcoptor is about two square feet and has a takeoff weight of less then three pounds.

You might be able to carry a small solar panel as a backup in case the copter runs out of power in a remote location, but it would have to sit for quite a while before it's ready to fly again.

Only in massive sizes, and only for planes (vs rotorcraft): https://www.theguardian.com/environment/2016/jul/26/solar-im... (something autonomous could of course be smaller, but still)

Drone with Apps & Games: http://adia.tech/

the idea is amazing, but is not available and the website is pretty horrible the subscribe link don't work and seem made by screenshoot..

The entire website is just one giant image.

With all the below, you'll have a solid drone with more power than you are likely to need even in heavy wind. You can program it in a couple ways. One way would be to use the flight controller provided, put it in air mode, and balance the signal with an external arduino/mpu6500 combo that feeds the signal to the normal receiver inputs. Another option is you can just go in to the BetaFlight code and modify it to your liking. A third option, is to remove the flight controller altogether and just use the arduino/mpu6050 combo to feed signal to the esc.

If you are trying to use the camera for some sort of remote processing, you'll need a vtx [7]

The thing no one is saying so far though, is be careful. Playing with your own balancing software on a quadcopter is an easy way to turn it into an unpredictable missile with an exposed blender attached. You will get hurt or someone else will get hurt. Do it inside and tether it down at first, that way the only day you ruin is your own

Here's a part list for you ($108 total):

- [1] Lisam-LS-210 Frame ($16): cheap, sturdy, lots of room for your stuff

- [2] Matek Power Distribution Board ($4): easy to run power to everything, plus it gives you 5v and 12v BEC for powering your other stuff like an Arduino

- [3] Omnibus Flight Controller ($25): runs cleanflight/betaflight open source flight controller software. That way you can choose whether you want to write your software on the controller or on your arduino

- [4] 4x 2205 2300KV brushless motors ($25): these will let you add a decent amount of stuff to your platform without it bogging down

- [5] 4x 20 amp esc's ($26): Needed for the brushless motors. They run BLHeli control software, so yet another thing you can fiddle with

- [6] camera ($12): simple camera, decent but not hd quality

[1] https://www.banggood.com/Lisam-LS-210-210mm-Carbon-Fiber-Fra...

[2] https://www.banggood.com/Matek-Systems-PDB-XT60-W-BEC-5V-12V...

[3] https://www.banggood.com/Betaflight-F4-Flight-Controller-STM...

[4]https://www.banggood.com/4X-Racerstar-2205-BR2205-2300KV-2-4...

[5] https://www.banggood.com/4-PCS-Racerstar-RS20A-Lites-20A-Blh...

[6] https://www.banggood.com/700TVL-2_8mm-Lens-100-Degree-Wide-A...

[7] http://www.banggood.com/TS5823S-Micro-VTX-5_8G-200mW-48CH-Mi...

All this is basically straight from this part list, but seems applicable