cfbridge.py vs cfbridge c++

[barzanisar]

Hi,
I am using your cfbridge.py that seems to work better (i.e. with less than 0.5% data loss) as compared to the cpp source code while communicating with crazyflie using crazyradio PA. The only modifications I made was to set the number of retries in radiodriver.py to 3 and to only send CRTP messages containing mavlink packets to QGC in here i.e. by checking if the pk.port== CRTP_PORT_MAVLINK

I will open a PR soon for these changes once bitcraze replies how I can set_arc using their crazyflie class and hence in cfbridge rather than hardcoding in radiodriver.py

Data loss while sending msgs from QGC to Crazyflie (via cfbridge c++) is around 5-7% , even with sending joystick commands at 5Hz. This seems to be the significant factor in creating issues with radio communication, such as arming/disarming command is not received by crazyflie. However, cfbridge.py works fine with the modifications discussed above, changes in syslink, and in QGC to send joystick messages at constant rate of 10Hz (will open a PR soon).

[barzanisar]

Solved:
To use cfbridge.py with CF2 flashed with PX4, I modified the details in PX4 user guide. I created a fork of crazyflie-lib-python that contains the updated cfbridge.py with number of retries set to 3. The necessary changes in the radiodriver.py are also in the fork.

[barzanisar]

However, the problem of data loss with C based cfbridge is still not solved.