Beaglebone Quadcopter!

Hey guys,

A group of Rose-Hulman students have been hard at work this past year building a Beaglebone Quadcopter with these goals in mind:

  1. Low cost ($100-150 w/o Beaglebone)
  2. Fully open source (Cape, Frame, and all control software) on our Github.
  3. Easy to assembly and repair (we estimate it will take 1-2 hours to assemble and get ready to fly)
  4. Durable and easy to fly (currently supports USB game controller control, though we initially tested using a USB RC controller, though it’s also possible to eventually use a cell phone as a controller) with a 10 minute flight time
  5. Sensor packed: 9-Axis (MPU 9150), Altimeter (BMP 180), Ultrasonic Rangefinder (HC SR04), Battery Gas Gauge (MAX 17044), and CMOS Camera (OV7670).

The Quadcopter is flown via WiFi and Bluetooth (though streaming video doesn’t yet work over Bluetooth) from a host computer (you need somewhere to view the streaming video from). Additionally, we’re using a Debian image and have added Xenomai for better real time performance. We’re also using both PRU’s: one for real time motor control and one for the camera. With the quadcopter software running, we’ve still got 80+% of the CPU free for other processing (OpenCV, etc.).

We’re currently using a PID control scheme, but we may be switching to a sweet state variable feedback system (or getting a senior design group next year to do it).

So, we want some feedback from you guys on the following:

  1. Would you buy one of these Quadcopters?

  2. Is our price point reasonable? Is this something worth selling ourselves or would this be a good kickstarter project?

  3. Are there any other features you think are critical (wouldn’t buy without it)?
    If you want to dive deeper into our design:

  4. How does the software look (particularly the PRU to C interface)? Would you be willing to maintain it, or update it to State Variable?

  5. How does the PCB look? Are there any flaws you see? Would you want to add or remove any other sensors?

  6. How does the mechanical design (Quadcopter frame) look? Is it aesthetically pleasing? Easy to build (if you have a laser cutter, give it a try and let us know)? Easy to fix when broken?
    Here are some pictures:

Videos are available on here (and on the github). Our best video is probably here, as it shows it both flying as well as the streaming video.

If you guys have any questions, respond here or email me back.

Thanks!

I would like a longer flight time and something that can be flown outside in some wind. My primary focus is delivering a BBB equipped with 802.15.4 and wifi running my pentesting Linux distro to hard to reach targets.

Your project looks excelent and it is by far the coolest project involving the BBB. I have a few BBB laying around from an old project (about 50 :slight_smile: ). Might consider trying something like this. Great work guys really great work.

I like it and for $100 I would buy one immediately, $150 would make me think twice. It would also be fun for me as a builder to add a traditional radio remote to it also.

Hi,

Your project certain looks really interesting .
Yes, I would be interested in buying one if the proce doesn’t exceed $250,-- but depending on the features
this would not be my limit.

Are you goinh to publich the diagrams and Software?

Can you post/email regularly.?

When do you expect a compeletion date?
Postage to the EU a problem?

All the best wishes for your project

NT

Mike,

That is a very interesting project. For ~$250 mark I think it is a viable kickstarter project. I’d buy one. I like the idea of having the extra compute power to experiment with OpenCV, sense and avoid systems, etc. Some things that would be nice to add / change (although these aren’t really deal breakers) are GPS position and altitude hold (perhaps using a uBlox NEO 6M.) One thing that does concern me is the frame. It looks like laser cut acrylic. I don’t think that material is strong or rigid enough for a quad frame of that size. I’d look into G10 as an option. This would increase your cost, but would provide a much better frame. Another option is to look at some of the cheaper, off the shelf frames available. Flite Test makes some great kits using mostly wood (and some G10) that are cheap and possibly more robust. They are also very easy to fix in the event of a crash. The H quad has the largest payload platform, you can find it here; http://shop.flitetest.com/multirotors/knuckle-h-quad-370-kit/ I think a robust, or easily repairable frame is very important considering the application here. Since this is such a great platform for experimenting with autonomous or semi-autonomous flight there are most certainly going to be crashes. Having a frame that will survive or be easily repaired would be a must.

Thanks,
SKip

A traditional radio transmitter can be used via the USB (similar to how RC Simulators use them). As for a receiver… We currently have control and video streaming over WiFi (or Bluetooth), so you’d have to implement a two radio solution. A WiFi or Bluetooth USB dongle would be included, or could be provided (would be cheaper to provide your own).

A note on pricing: the quadcopter parts kit (everything except the BeagleBoard) would be sold for around $100. This doesn’t include a controller (to keep the cost down, we can’t). The thought on this was that most people have access to a USB controller of some sort (Xbox, etc.).

–Mike

The project is feature complete as of now, and everything is available on the Github:

Were we to sell a product, it would likely be 6 or so months before we would begin shipping everything, as we would have to get parts pipelines set up, production going, and good fulfillment going. That being said, if you wanted to, we’re confident you could build your own (assuming you have access to a laser cutter [or a steady hand] and a way to make PCB’s [probably a low volume quick turn place like Advanced Circuits or Silver Circuits {in the US at least, not sure what’s best for where you are}]).

Frame CAD Designs: https://github.com/Rose-Hulman-ROBO4xx/1314-BeagleBone-Quadcopter/tree/master_rev2/noncode/CAD

PCB Design: https://github.com/Rose-Hulman-ROBO4xx/1314-BeagleBone-Quadcopter/tree/master_rev2/noncode/Board

Software: https://github.com/Rose-Hulman-ROBO4xx/1314-BeagleBone-Quadcopter/tree/master_rev2/code

We will be publishing a .img file with our custom Debian image as well, so it can be easily flashed.

We don’t know what our plans are for shipping (especially overseas, we may have issues with US Export laws concerning shipping quadcopters with GPS capability [ITAR]).

Thanks for your interest!

–Mike

Skip,

Thanks for the feedback. I forgot to mention above, but we’re including a GPS on the Rev 3.2 boards (just got them and will assemble them soon) that’s the same as the Adafruit Ultimate GPS. It seemed to be a good combo of power and cost (we can get them in bulk for $15). We haven’t incorporated this into the code, but given the nature of OSS, we’re taking the Field of Dreams approach (“If you build it they will develop with it”), providing a platform that works and can be expanded upon as more users desire that functionality.

Version 1.0 of the frame was laser cut acrylic, and you’re right, it flexed a good bit (which worried us a lot). We also made it out of wood for comparison and got significantly better performance, and slightly less flex (for half the weight). We considered materials like G-10 (actually just making the cape the frame [terrible idea since if it breaks you’re screwed and have to replace the most expensive part]) and carbon fiber (expensive and difficult to work with as laser cutters have issues with carbon), and settled on this frame design since it provides the minimum amount of structure necessary to provide what we need at the given cost point. We have also included an extra arm and extra landing plate in case either of these break (possibly just two arms in the future). The design we have also allows for adding longer arms (and more arms) to easily make any geometry multicopter (tri to octo), while a “traditional” frame simply couldn’t do that.

As for durability, we’re a bunch of college kids, so breaking things is pretty much routine. We’ve spent a good part of the last month crashing this quadcopter into the ground, the ceiling, walls, ourselves, trees, etc. and seem to be getting pretty good durability from it. We could have made the frame out of a stronger material (G-10, aluminum, etc.), but we actually kinda want the frame to break before anything else on the quadcopter does, since the frame is the cheapest and easiest to repair, we’d rather have it fail than have a strong frame end up transferring force to the cape or the bone. I think the easiest solution to making our frame more durable would be to double up on the arms (put a set on the top and bottom of the main plate, and beef up the supports by where they attach) to get some 3D strength (like the frame you posted). I think we will try a stiffer material (G-10 or similar fiberglass) and see how that goes. I could even see a world where we simply make the base frame and the arms as PCB’s and run traces for the motors out to the end of the arms.

Thanks and let me know if you have other thoughts around this,
–Mike

I just checked G-10 prices on McMaster, and they are a bit steep for what we’d want to sell (about $20 for a 1/8"x1’x1’ sheet) (compared to just over ~$1 for the 1’x1’ sheet of plywood).

I think we’re going to end up with several tiers of quadcopter frame, with the base model being wood or acrylic and higher end models using G-10 or Carbon (or aluminum, possibly though that’s ~30% heavier than G-10).

The quadcopter currently weighs about 400-450 grams, and we’re generating around 800 grams of thurst, so we could use heavier frames, it just degrades performance some.

Lastly, the CAD designs are all out there, so you’re free to make one out of any material (exotic or otherwise) that you can machine :slight_smile: Personally I’d love to see the titanium quadcopter!

–Mike

Hey guys,

A group of Rose-Hulman students have been hard at work this past year building a Beaglebone Quadcopter with these goals in mind:

  1. Low cost ($100-150 w/o Beaglebone)
  2. Fully open source (Cape, Frame, and all control software) on our Github.
  3. Easy to assembly and repair (we estimate it will take 1-2 hours to assemble and get ready to fly)
  4. Durable and easy to fly (currently supports USB game controller control, though we initially tested using a USB RC controller, though it’s also possible to eventually use a cell phone as a controller) with a 10 minute flight time
  5. Sensor packed: 9-Axis (MPU 9150), Altimeter (BMP 180), Ultrasonic Rangefinder (HC SR04), Battery Gas Gauge (MAX 17044), and CMOS Camera (OV7670).

The Quadcopter is flown via WiFi and Bluetooth (though streaming video doesn’t yet work over Bluetooth) from a host computer (you need somewhere to view the streaming video from). Additionally, we’re using a Debian image and have added Xenomai for better real time performance. We’re also using both PRU’s: one for real time motor control and one for the camera. With the quadcopter software running, we’ve still got 80+% of the CPU free for other processing (OpenCV, etc.).

We’re currently using a PID control scheme, but we may be switching to a sweet state variable feedback system (or getting a senior design group next year to do it).

So, we want some feedback from you guys on the following:

  1. Would you buy one of these Quadcopters?

Absolutely!

  1. Is our price point reasonable? Is this something worth selling ourselves or would this be a good kickstarter project?

$100 is reasonable. $150 is kinda pushing the limit a bit for me. Looks great for experimenting and I think it’d be a good kickstarter. Hopefully you build up some good community interest here first and discover the killer feature for which everyone needs this. :slight_smile:

  1. Are there any other features you think are critical (wouldn’t buy without it)?

Having the camera is awesome. I might want an optional GPS and/or 3G modem. Can I add one myself? I don’t think it is required in the bundle.

If you want to dive deeper into our design:

  1. How does the software look (particularly the PRU to C interface)? Would you be willing to maintain it, or update it to State Variable?

Looks like the main routine is control_alg. Personally, I kinda like the structure of libpruio from what I’ve seen so far, where there is a C library interface provided. I also like the messaging approach of Matt Ranostay’s PRU lighting code. Do you have any code documentation that would make it easier to review? I don’t think many people are going to dive into your code until they know what problems it solves and how modular it is such that they feel they could start using parts of it.

Looks like you have a lot running on the PRU. Do you have a summary?

  1. How does the PCB look? Are there any flaws you see? Would you want to add or remove any other sensors?

Layout seems pretty simple and clean. Is 1314-BeagleBone-Quadcopter/Quad32brd.png at master_rev2 · Rose-Hulman-ROBO4xx/1314-BeagleBone-Quadcopter · GitHub the latest? All of the versions are a bit confusing for me.

  1. How does the mechanical design (Quadcopter frame) look? Is it aesthetically pleasing? Easy to build (if you have a laser cutter, give it a try and let us know)? Easy to fix when broken?

I have access to a laser cutter at my hackerspace, but I’ve never used it. Do you have simple pictures of the pieces or any hints on where I could order some pre-cut pieces like with Shapeways?

As long as you keep adding unnecessary stuff to this thing it gets less time for fun (flying) because 3G or other dongles consume at lot of current. They are just designed to be fed by a wall outlet

I have some experience with 3g dongles on the beaglebone and they are very very unreliable. I’ve tried different models and they all suck. Also they are powerhungry, about 300 - 600 mAh at 5v. Depending on the state they are in idle or high power

Mike,

The project sounds interesting.

I would recommend partnering with someone like Adafruit, Sparkfun, or Makershed. There are about two million things which you need to get correct to successfully being a new product to market. Logistics are a nightmare.

As collage student working in this market you are probably better off focusing on building your reputation rather than focusing on the money.

See if you can get beaglebone.org of makezine to publish a series of tutorials on the project of do a set of youtube howto videos yourself. That way you can refine the product and get market feedback while building your reputation.

At this point, I would be happy to purchase a PCB, frame and parts list as a DIY project :slight_smile:

Hey,

The project is definitely interesting. I would definitely buy one to play with.

Keep me posted

Felix

This sounds GREAT! I’ve just started trying to build the very same thing just on a whim so for anything under $200 I’d be all in! For $150 it would be really hard for me to only buy one.

If its a Kickstarter Project you’ve DEFINITELY got a backer here.

Please, keep me posted!

-J.B.

Be sure to check out beagleros.wordpress.com
about the same subject, and an autopilot!

Russ

John Harris (john.customstage@comcast.net) I would like to build on of these capes if I can buy a bare PCB. Are there any available?
John

hye Mike MCDonald
I would like to know what is the name of the board that you connect with BBB…the green one board…
I hope you will reply my email…thank you…