Bitcraze Forums

We are working on an autonomous swarm and I had some trouble where drones seem to hang when certain commands where issued on the drone.
Whenever I wanted to debug via radio / usb connection and the console prints, everything was fine. So this took me a while...

Turns out I was writing parameters within my function. When parameters are changed these are reported back to the connected client. This function blocks! If there is no client it will wait till there is one.

Is there a reason the clientupdate is implemented blocking? Also found the compiler switch to deactivate this (define ), still I do not see the reasoning for the blocking.

What I am trying to do in the first place is setting the state estimation to a new value (e.g. not providing a meassurement but actually overwriting the internal state). Is there a nicer function than setting the parameters and thereafter ?

( PS: it would be nice to have a way to have "inline" BBCode for fixed width fonts to refer to functions in text rather then blocks. )

I do not think this function was meant to be blocking. What is happening is that when you set a parameter from inside the Crazyflie, it sends a packet to the client with the new value. If no client is connected, the packet TX queues become full and so the function blocks until it can put its packet in the queue. This is to keep the original assumption about parameters: they can be read once by the client and they are not changing dynamically in runtime. This packet allows the client to still keep track of changing parameter.

So, for your use-case, the SILENT_PARAM_UPDATES build flag seems like a good one to use. Another solution would be to not send these packet if no client is connected, they are only needed if someone is connected over CRTP anyway so I think this would be an OK behavior.