Ground Truth for Sim Backend

[mwlock]

Hey all, for the simulation backend, I am wondering if there is a mechanism for getting the state of the UAV? For example, after stepping one could publish an Odometry message. Perhaps at a lower rate.

crazyswarm2/crazyflie_sim/crazyflie_sim/backend/np.py

Line 37 in 379601e

uav.step(action, self.dt)

If this does not exist and might be useful, I would be happy to create something simple and make a PR 😄

[whoenig]

Where do you want to get/query the state? Each visualization has access to it, see https://github.com/IMRCLab/crazyswarm2/tree/main/crazyflie_sim/crazyflie_sim/visualization. For example, one can write the state(s) to a CSV file for later plotting. A new visualization could be written for publishing a ROS message, but then it would make most sense to use the /pose message that is also used in the real drone (if logging is enabled).

[mwlock]

I am looking to subscribe to a ROS topic to mimic the MOCAP system/state estimation as per your second suggestion. I have yet to perform experimentation on the real platform and am looking to test in sim first, hence why I am looking for the ground truth

[whoenig]

In that case this would be simulating the motion_capture_ros package, which publishes on /pose as well as via tf. Another option would be /<robotname>/pose, which would simulate the behavior when using the on-board estimated state from the EKF rather than the global mocap state.