August 24 2024: https://www.reddit.com/r/AerospaceEngineering/comments/1f0h9p6/i_invented_a_winch_to_control_kites_eventually/
August 24 2024: https://www.reddit.com/r/boating/comments/1f0ii4q/i_invented_a_winch_to_convert_boats_into/
January 29 2024: https://www.reddit.com/r/3Dprinting/comments/1adupcm/ive_been_secretly_3d_printing_this_bot_that/
It’s used for home automation primarily, but the OG use for MQTT was oil companies needing low power long range devices to collect telemetry on oil pipelines.
This is an excellent primer on MQTT: https://www.youtube.com/watch?v=jTeJxQFD8Ak&list=PLRkdoPznE1EMXLW6XoYLGd4uUaB6wB0wd
The thinking is you put your sensor package into a kit with some solar panels and batteries so you can drop it on a pipeline in the middle of the wilderness and it constantly transmits telemetry over a packet radio transceiver. There is a hub that is receiving all of the telemetry and distributing the data to all of the subscribers.
So, in this case you’d have accelerometers, anemometers, or you can even get mm-wave radar range finders (which may or may not work on the water’s surface YMMV) all attached to something like a Raspberry Pi. The Pi would transmit the telemetry via MQTT over a LoRa WAN to a base unit running the MQTT broker software and then any applications that you would be running locally (like the control software) could receive a stream of real time telemetry updates from the onboard sensors. The MQTT broker is just a service that you run, but it can be (in giant deployments) its own device if required.
Here’s an example of someone setting up MQTT to transmit over LORA, they’re sending different data but the process would be the same but you’d be reading from different sensors: Connecting long range sensors using LoRa and MQTT - Share your Projects! - Home Assistant Community
Depending on the expected flight time you can probably run it on batteries for quite some time. The battery weight should be trivial compared to the amount of lift the kite is generating… so even if you needed a few kg of batteries it shouldn’t (? I’m no kite engineer) affect the kite handing too much. Alternatively, you could run cable up the control lines to carry power and data, though the total weight for something like this could become a factor… but you wouldn’t have to worry about issues with wireless causing you to lose telemetry if it becomes a critical part of the control system.
All of the components in the Arduino ecosystem are built from the ground up so there isn’t a lot of excess components to add weight. Batteries would be most of the mass, the sensors and compute are very light.
[–]photoshopbot_01
Yeah, it could be built under 100g easily. I make small hobby remote control fighting robots (with a weight limit of 150g including all the motors, wheels etc) and I experimented with adding an mpu6050 to my setup. stick it in a mini tupperware or other hard sealable case, and maybe something printed in tpu on the outside for a bit of shock absorbsion and you should be able to throw it at a brick wall as hard as you like without breaking it. I use an arduino with integrated NRF24L01 (called an rf-nano) but it’s only super reliable for 10-20 metres or so. For your application I think I would try something intended for longer range stuff like LoRa.
I love this. If you printed a large 2-D barcode on the kite, you can get the exact position and angle with a camera system, and use a controller to automate it.
Furthermore, if the system knows the speed of the boat, the angle of the kite on the force of the tug, you could use it to get wind, speed data.
If the barcode is reflective, you could illuminate it with just a low energy laser.
That same laser scanner could project letters and numbers onto the kite at night identifying the vessel.
Very tempted. I wrote an SBIR proposal for something similar about 13(?)years ago. It was for forward outposts. The cords were specced similar to DX50 antenna wire for radio. There was talk of sending up a ducted fan drones up the cables for automatic launch and extended loitering.
We used fiducials for our robots. Cheap and accurate positioning.
Related:
Using cameras to see an object and some other stuff: Real-life AIM-ASSIST: Aimbot V3