I’ve combined this all together into a custom carrier board jointly developed with Running Electronics but for those hoping to do something similar, this AP/BlueOS setup page has mostly equivalent off-the-shelf parts.
What most excites me about this vehicle is that the software is nearly completely open source from top to bottom. The “bottom” being the AM32 DroneCAN ESCs which support control, feedback and OTA over DroneCAN and the “top” being BlueOS which I’ve blogged about before. This gives us the ability to properly integrate everything into an easy-to-use and highly functional package.
On this copter we’re using BlueOS in a number of ways:
The vehicle is not complete yet, and is not for sale (yet). We’ve still got two serious issues to resolve including GPS interference (perhaps from the USB connection between the RPI and LTE modem) and high vibration (perhaps the propellers pass too close to the frame) so I’ll provide more updates in the future.
Until then here’s a video of the vehicle flying in my garden
If you’d like to get involved there here’s some things I’m looking for:
Any advice on the frame, hardware, etc is greatly appreciated
If you’re a camera manufacturer, the community needs more small cameras that are ethernet enabled and ideally allow us to control the image quality and frame rate
If you’re an autopilot manufacturer, the community needs more ethernet enabled autopilots
As you may already know T-Motor and the likes are banned for import into the US. I understand that many other countries can continue to use them as a vendor, but having a non T-Motor variant for propulsion would be nice for folks in the US (if you plan to sell to the US).
Also - for what is worth I think having this be a folding frame would help a lot with transport/portability. I also think that if you optimize it well you can get it down to running a 4s 5Ah pack of a more standard cell (say Molicel P50B) with an all up weight of around 1000g and 10" propellers while still maintaining 30+ minute flight times w/ a 20% battery reserve.
Example:
900-1000kv Motor w/ 10" Propeller
4s 5Ah 21700 Li-Ion Battery
Lightweight Airframe
Lightweight Camera
Target AUW of 1100-1200g leads to 30+ minute hover time w/ 20% battery reserve.
Category
Weight (g)
Airframe
250
Propulsion
300
Autopilot and Electronics
200
Battery
300
Payload/Camera
75
Misc.
25
Bigger props are obviously more efficient and will get you more flight time. So if you want to run a similar setup with slightly heavier / lower KV motors and 12" propellers - that works to. Just thinking of ways to make the aircraft as cheap and compact as possible.
Yes, I’m working with VimDrones to develop an open source (software) BMS. The development hardware is already done so it’s just a matter of writing the software. I can’t promise when that will be complete but hopefully around the end of the year.
Dear rmackay9 If you want to build something which is needed. It should be better in terms of quality of photo and video from standard drones like Mavic. So please consider professional camera like Sony ILX-LR1
Here’s a little end-of-November update on progress.
above is the video of Leonard flying the vehicle at the AP dev conference in Richmond
Somewhat surprisingly the GPS issue appears to be a power issue rather than an EMI issue. During the investigation we used UBlox’s built-in MON/SPAN message and we did see some interference from the carrier board but the larger issue appears to be that the carrier board (being developed with Running Electronics) was powering all the CAN and serial ports from the Cube’s 5V peripheral power pin which only provides up to 1amp which is apparently too little to power the Here4 and 2x benewake lidar at least during startup. In the next version of the board, the Here4 will be powered all on its own by the Cube’s “high power” 5V and the lidar will be powered by the board’s main 5V/10amp regulator
The BMS dev board (being developed with VimDrones) used an SPI connection to the TI-BQ76952 chip but we found it was quite difficult to work with so a new dev board is being produced that uses I2C
Re 3D printing, one of our prototype’s arms broke during vibration testing. Based on some great advice from a Japanese company called Boduk we think that our bamboo printer’s carbon fibre print doesn’t have enough vertical strength so we’re going to get some professional prints done using MFJ (mult-jet-fusion) and PA12 Nylon.
An ArduPilot community member provided a contact at a Taiwanese camera manufacturer and it looks like I may be able to get a camera that fits within an XFRobot or CADDX gimbal but also connects directly to the RPI CM companion computer. This would be a huge advance because it would reduce latency and give us full control over the camera. This is still very much in the early prototype stages though so no promises!
Thanks very much for the comments and advice above, I’m of course reading every word of it!
Regarding GPS interference, I’ve had quite a lot of experience with it, because I use “companion computers” in both the rover and the drone.
The biggest source of interference coming from these sources are the “pcie lines”, which are used to connect various interfaces, in my case it was mainly USB3 ports. Since the rover uses a 2xUSB ethernet adapter, I had to equip both adapters with the most perfect additional shielding possible using a galvanized sheet metal box.And I separated the “AGX Orin Devkit” computer with a duralumin board, against the “Cube Orange+DroneNet+Here4” set.
The result is that if “AGX Orin” is turned off I have inside approx. 8 satcount, when I turn on “AGX Orin” it drops to 4 satcount and less.
Outside I have approx. 18 satcount with “AGX Orin” turned off and approx. 14 satcounts when turned on, just yesterday I tested it again, because I started MAVROS & DDS XRCE over ethernet, as services for ROS2.
As for the problem with materials for functional 3D printing, I only use PC Max from POLYMAKER, all other variants including ASA, Nylon, CF etc., had significant problems with durability, hardness and overall resistance.
I had to build a larger 3D Delta printer according to my design “700mm height, 400mm printing diameter” with a heated chamber at 60 degrees Celsius and especially let the finished print cool down as gradually as possible to eliminate the natural internal tension of the material.
Without a heated chamber, there is no point in even trying …
As someone that has tons tons of testing with GNSS noise and other RF properties. I will highly state that any High Bandwidth USB 3 devices will cause tons of interference. Many GNSS receivers may even detect jamming.
You can attempt to add shielding, but without proper testing it can be a shot in the dark.
I’ve passed on the recommendation of PC Max Polymaker to my frame designer (Tsuyoshi Arakawa) and we will likely give that a try (we’re trying all kinds of things to see what works and what doesn’t).
The comment about a heated chamber is also very interesting. We suspected that the change in the weather could be the cause of the easily broken arms so a chamber could be the solution although putting everything in a chamber is more easily said than done!
I’m tempted to create a wiki page and/or discussion topic to allow more sharing of advice on how to 3D printed a frame.
I agree that heating the chamber is definitely not a simple problem, it took me quite a while to design and test it, but the result is great.
I can print really big things without tearing and twisting, especially with much greater durability in all parameters.
When I was looking for a ready-made commercial 3D printer to meet my expectations, I couldn’t find one .
So I sat down and started calculating, drawing, buying materials and manufacturing .
After several not-so-great experiments, I found that it is advantageous to ensure controlled circulation of heated air, when I heat the air in the upper part to 75 degrees C and send it down with 3 fans against the heated bed at 100 degrees C. The air that accumulates under the heating pad, on the other hand, I send 3pcs fans up towards the upper part of the chamber.
I have the control sensor for heating regulation at the level of the heated bed, at the peripheral edge of the printing space.
The print head does not have a fan, it is water-cooled with a working temperature of up to 500°C.
The source of heating the hot air is 3pcs "
200 Watt DC 12 V Electric PTC Heater Thermostat Insulated Ceramic Fan Heater High Power PTC Heating Element" powered by a separate switched power supply.
This is the first contribution to the new section “trouble with 3D print” …
Impressive Randy! I can recommend inserting a CF spar inside the arm that can take a hit or two . See photo. Basically I’m not under the impression that any drone is able to handle a hard crash from 4 m. A replacement part concept could be a solution.
I’ve come to the conclusion that HP’s MJF (Multi Jet Fusion) using Nylon PA12 is probably the best way forward. The core reason is that, similar to SLS (Selective Laser Sintering), the bonds it produces are equally strong in all directions (e.g horizontally and vertically). The prints are also extremely high quality, strong and rigid which means end users should be satisfied. FDM prints on the other hand have relatively weak vertical bonds (e.g. bonds between layers) and this is especially bad when using carbon fiber prints which is what we’ve been using (the Bambu Labs X1C). If you print the drone arms horizontally they’re fairly strong but end up with fairly rough surfaces which end users won’t like. MJF is of course the most expensive method of 3D printing but you get what you pay for.
The battery BMS software is progressing well (PR is here) although it’s still too early to tell when the BMS as a whole will be complete
FDM objects have to be designed with layer separation in mind and accommodate for it by either incorporating internal structures and the best way as @Quadzilla mentioned use some CF spar either tube or rods or even post tension cables vertical to the layers!
Voron Design FDM printers are designed for printing ASA, ABS, PA and other materials requiring heated chambers. Most of the Voron builders do not have active heated chambers, but by running the heat bead for 20 to 60 minutes, depending on the size of the airtight printer chamber, and using internal fans/carbon filters for VOC to circulate and filter the chamber air we can maintain about 50 to 60 C temperatures. Btw you can measure the chamber temperature using the thermocouple of the extruder, as long as the extruder heater is powered-off, by “parking” the toolhead in the middle of the print chamber.
Finally, we built the Voron printers using ASA and ABS filaments, no CF, using 0.4 nozzle that can operate almost non-stop with almost zero “plastic” part failures and or deformation in a ~55 C environment and achieve 500 mm/s speeds and 10,000 mm/s/s acceleration for years.
Of course, I would love to have access to a MJF printer, or it is a commercial product.
In case there was any doubt, the prototype we flew at the AP dev conference broke during shipping at exactly the same place on the arms as the follow-up prototype. 90 deg angles in 3D printed parts is not a good idea :-). A ground-up re-design has already begun!
.
…
