Open Source Motion Capture for Autonomous Drones (2023)
joshuabird.com
Open Source Motion Capture for Autonomous Drones (2023)
1–9 of 9 posts
Re: Open Source Motion Capture for Autonomous Drones (2023)
#2A big point of the article is about "open source" but no link to the github repo? Am I being blind? I scanned the article twice, cannot find it!
Re: Open Source Motion Capture for Autonomous Drones (2023)
#3A big point of the article is about "open source" but no link to the github repo? Am I being blind? I scanned the article twice, cannot find it!
Found it in the Youtube video description: https://github.com/jyjblrd/Mocap-Drones
Re: Open Source Motion Capture for Autonomous Drones (2023)
#4Not only well done but well-presented. Lots to learn from that codebase! I think the most impressive part is keeping track of which LED is which, especially with multiple drones flying around. I would've tried something goofy like PWM'ing them to make identification easier, but that would have added more latency than doing it the hard way with the math.
I wonder if the LED support wires are stiff enough to avoid some stability loss. Seems that any flexing could make tuning more difficult.
Re: Open Source Motion Capture for Autonomous Drones (2023)
#5"I just bought 8 cameras for $1 each, and 5 of them ended up being the correct type."
Monte Carlo Ebay Purchasing. I sense a good paper here.
Re: Open Source Motion Capture for Autonomous Drones (2023)
#6I’m more impressed with repurposing some parts like the cameras, good work!
So the commands were sent through RC channel, what was the bandwidth? Maybe I missed it but I didn’t hear it in the video.
Project looks great but the application is kind of limited as it relies on the external components (cameras), really hard to do in the wild, even in urban environments.
Re: Open Source Motion Capture for Autonomous Drones (2023)
#7I’m more impressed with repurposing some parts like the cameras, good work! So the commands were sent through RC channel, what was the bandwidth? Maybe I missed it but I didn’t hear it in the video. Project looks great but the application is kind of limited as it relies on the external components (cameras), really hard to do in the wild, even in urban environments.
> So the commands were sent through RC channel, what was the bandwidth?
It was over wifi, with the ESP32.
https://docs.espressif.com/projects/esp-idf/en/latest/esp32/...
Seems people say real world around 8mbps:
https://www.reddit.com/r/esp32/comments/9uh3l8/whats_the_esp...
Re: Open Source Motion Capture for Autonomous Drones (2023)
#8I’m more impressed with repurposing some parts like the cameras, good work! So the commands were sent through RC channel, what was the bandwidth? Maybe I missed it but I didn’t hear it in the video. Project looks great but the application is kind of limited as it relies on the external components (cameras), really hard to do in the wild, even in urban environments.
> So the commands were sent through RC channel, what was the bandwidth? It was over wifi, with the ESP32. https://docs.espressif.com/projects/esp-idf/en/latest/esp32/... Seems people say real world around 8mbps: https://www.reddit.com/r/esp32/comments/9uh3l8/whats_the_esp...
Thanks, that makes sense, I thought it was the typical ~900MHz which would be even more interesting.
Re: Open Source Motion Capture for Autonomous Drones (2023)
#9I built something similar and the position accuracy is really great.
However, I only used a single camera mounted directly below looking upwards.
I estimated the distance from the perceived distance between the LEDs (assuming the drone is flying slowly and parallel to the ground).
Using this setup I implemented a controller that would land the drone "on" the camera. And this way much better landing precision than using GPS can be achieved.