How to use odometry and cmd_full_state at the same time

[allegorywrite]

I'm currently trying to integrate a planner that subscribes to odometry and publishes cmd_full_state, into crazyswarm2 .
However, the document says that cmd_full_state and odom functions are independent in cpp and python respectively, and I cannot use both at the same time.
Please provide current information regarding this issue.
If it is not currently available, I would like to customize crazyflie_server. Is it technically possible to use both cmd_full_state and odom?
If possible, please let me know which is easier to modify, cpp or cflib.
I'm also considering using crazyswarm1 in some cases. However, when I checked, it seems that odometry is not provided by default in crazyswarm1.
Thank you for your answer in advance.

[knmcguire]

That is indeed a bit of a dillema! You are correct on that for sure.

The full state command wasn't available in the crazyflie python library yet until recently with this pull request: bitcraze/crazyflie-lib-python#382. An example script in python can be found here: https://github.com/bitcraze/crazyflie-lib-python/blob/master/examples/autonomy/full_state_setpoint_demo.py

You can implement it yourself perhaps easier in python but I'll give a go on a quick PR that I'll hopefully be able to finish this afternoon.

[knmcguire]

Here is the PR #350

and this is the branch: https://github.com/IMRCLab/crazyswarm2/tree/cflib-full-state-cmd

The reason why it wasn't implemented (and why we should wait to merge this) is that this is not yet part of the latest cflib release but of the latest development branch. you should install cflib with source: https://www.bitcraze.io/documentation/repository/crazyflie-lib-python/master/installation/install/#install-from-source

[allegorywrite]

thank you! That helps a lot.
However, it seems that cmd_full_state is still not perfect.
Two msg types are defined in the cmd_full_state topic as shown below.

ros2 topic echo /cf231/cmd_full_state
Cannot echo topic '/cf231/cmd_full_state', as it contains more than one type: [crazyflie_interfaces/msg/FullState, crazyflie_interfaces/msg/Hover]

Personally, it seems like the cause is using hover for cmd_full_state in the PR commit.
I made the appropriate corrections and issued a PR.

[knmcguire]

oh yeah absolutely that would cause it. Classic copy past issue, as I made this PR purely in the github website editing screen. I kept in in draft since this code wasn't flight tested yet so I was planning to do it today 😄 but happy you did! At least I was able to give you a head start of how to do it.

Thanks for the PR ! I'll merge it since it is in a seperate branch anyway, and once we (at bitcraze) get the november release out for the crazyflie python library, we can merge it completely.

[knmcguire]

You can also try to implement it through the cpp route, but the problem there is that I'm not sure if the crazyflie logging type of FP16 (16 bit float ) has been implemented in the logging block initialization of the cpp backend. @whoenig, would you know about that by any chance?

[whoenig]

No support for FP16, yet:-(

[knmcguire]

ahhh... yeah oke then the cpp route is currently not possible then for now. The 'FP16' is essential for this. @allegorywrite, you should go for the python route as I described in the other thread