Crazyflie ros cmd position and external pose

[taizoonaliasgar]

Dear Dr. Hoenig,
I am using crazyflie__ros for waypoint following with Aruco markers. Going by one of your posts where you suggest using onboard controllers for position control instead of crazyflie_controller, I have written ros nodes that publish to topics cmd_position(target position) and external_pose(current position and orientation) subscribed by crazyflie_server.
When I try to run the launch file the rotors spin but the crazyflie does not take off. Is this enough information for the onboard controller for waypoint following?

Thank you.

[whoenig]

Yes, this information should suffice to achieve flight. Which firmware are you using? I assume cmd_position and external_position are different, i.e., the on-board controller has a position error to compensate for?

[taizoonaliasgar]

Thank you for your response Hoenig.
I have the latest firmware on the crazyflie, 2019.02 and am trying to correct for position error given current position(published to external_position) and setpoint(published to cmd_position)

I have attached an image of the launch file here. In a nutshell, I am publishing to the two topics from the two nodes talker and pose.

Do I need to setup the Kalman filter parameters like in position.py(attached here):

I will be glad to provide any other information you require.

Thank you.

[taizoonaliasgar]

Dear Dr. Hoenig,
I edited the config.mk file in firmware to include ESTIMATOR=kalman. The crazyflie takesoff at high speed to crash. I have setup kalman resetEstimation to 0 going by the Position.py script above.
Do I need to setup the update params service? Could you please help me figure what I am missing?
Thank you.

[whoenig]

• You don't need to recompile the firmware. You can force the Kalman estimator by setting "stabilizer.estimator" (a CF parameter) to 2
• You should initialize and reset the kalman filter, similar to https://github.com/whoenig/crazyflie_ros/blob/master/crazyflie_demo/scripts/publish_external_position_vicon.py. Note this assumes that yaw is 0. I believe newer firmware versions also have a parameter to initialize yaw.