Hi all,
I’m working on a visual precision landing problem. The initial approach was to have a visual marker-based pose estimator running in the companion computer and passing this information to the FCU which then handles the precision landing on the visual marker.
This discussion is a continuation of an issue I created in GitHub>>Mavlink>>Mavros>>Issues. This is the link for it. Having poor results with the Arducopter Precision Landing, which we interacted via landing_target plugin, we moved on to the PX4 Precision Landing.
Reading this website, it’s my understanding that the Precision Landing in PX4 is only compatible with IR Lock type precision landing. Speaking with Jaeyoung Lim (@Jaeyoung-Lim), I became aware of the Safe Landing Planner. For the past two weeks my colleague Nazar and I have been working on the safe landing planner. As Nazar stated:
I went through with the documentation. It is working on the output of a stereo camera. The output will be in 3D point cloud which will be converted to the binary polar histogram. This histogram data will be published to the local planner. The local planner used the min. value from the binary polar histogram and sends the instructions to the FCU to move the drone to that value.
We have a few questions about safe landing planner:
- In the simulation testing of the safe landing planner, the tf_depth_camera node is not publishing any data due to which the error - “avoidance system not ready” showing on the console.
- How does the local planner node handles the binary polar histograms from the stereo output ?
- Both local and safe landing planner are not working when enabling the com_obs_avoid param.