@tom, when that happens, does the LED do the "blink-blink-pause-blink-blink-pause..." sequence, i.e. the sequence listed here under "Binding mode enabled": https://www.expresslrs.org/quick-start/led-status/.
That's the same mode I get when I do voxl-elrs --bind, so is there a difference between doing it via the command vs doing it by holding down that button?
@tom, just checking to see if you had any further thoughts on this.
@Jainik-Mehta I have a Commando controller, and yes, all the LEDs on the drone operate as expected
@tom, one more question, we're trying to run /usr/share/modalai/voxl-elrs/tools/debug.py to try to see the logs, but the code doesn't seem to work.
For example, it looks for dev.port = '/dev/tty-slpi-7', but we can't find that file, so is this debug.py actually the right code to use?
Also, this error message is returned, even on a drone that is working correctly that binds to the controller.
ERROR: voxl_uart_read_bytes: Bus '17' is not initialized
PS, regarding the original post, unbinding and flashing did work, we can see the LED changes, etc, and messages from the drone
@tom, in other words, aside from these docs https://docs.modalai.com/voxl-elrs/, where are these instructions that refer to using a button on the ELRS receiver to bind?
@tom, can you elaborate which button on the ELRS receiver you are talking about? You mean the power on/off cycling button? Is there a way to press it such that it binds to the controller?
I'm not able anymore to bind the controller that comes with the Starling V2.
I've done multiple power cycles of the drone and controller, and ran the voxl-elrs --unbind and voxl-elrs --bind commands and go through the binding process, which used to work to rebind the joystick, but I'm no longer able to.
I then tried to reflash the receiver as per https://docs.modalai.com/voxl-elrs/#recovering-a-bricked-receiver, but still can't get the joystick to bind.
Any further suggestions to try?
Hello, has anyone tried to set up PyTorch on the Starling 2 drone?
I know there are TFLite models available, but I have a YOLOv8 model I've fine-tuned (using Ultralytics) and have converted it to a TFLite model, and it's been quite challenging to get it to work on the drone.
So, I was wondering if a better way is to somehow get PyTorch installed on the drone and use that instead for inferencing, instead of TFLite.
I know there's this new ExecuTorch thing available that enables inferencing using PyTorch models, so that I don't need to convert to TFLite:
I'm more of a data scientist and not a software engineer, and certainly not well-versed with embedded systems, so was wondering if anyone has set up ExecuTorch on a Starling 2 drone.