I’m having difficulty achieving basic hovering and position control with Pixhawk v2 and PX4 flight stack, despite having high quality optical flow data and motion capture feedback from a Vicon camera array.
Achieving reliable position stabilization is a prerequisite for a very exciting project that I am working on which is related to chemical localization. The PX4 flight stack is attractive to my research group because of its open source nature and uses ARM + linux. Currently our project is restricted to indoor environments where GPS is not accessible.
I would like to get in touch with someone who has successfully demonstrated position hold or position controlled flight in an indoor environment. I don’t know if I have a tuning issue, an integration problem, or if my airframe is unreliable. Anything that anyone of you can do to help me move this project forward would be immensely helpful.
I fly almost every day, and I am saving all of my flight logs for analysis; there is plenty of data that I can share to help address the issue.
If you can help or would like to know more about this project please contact me at email@example.com or simply message me via the PX4 discuss forums.
Same here. Cannot get it fly stable with optical flow only i’ve tested it in simulation with SITL and it seems to be a litte bit unreliable. even in perfect conditions like in a sim enviroment
@TobiasM Can you share a screenshot how you test? Because you would need actual texture…
@LorenzMeier i have used the standard “asphalt ground” in gazebo and a texture of our floor looking like this but in green color: http://www.teppichboden-pvc.de/pvc/pvc-bilder/inver_intermezzo/intermezzo_242.JPG
what i did:
- started with gamecontroller
- waiting for flow to resume
- setting pos.hold
-> sometimes it is working quite well with a little bit drift. but sometimes it says goodbye and flies in some direction. thats why i call it unreliable.
and in reallife you can completely forget it. i chose gras as texture but it goes crazy when i set it to pos.hold. the flow is focused on 2m distance. my next step is to attach a 1D lidar. which is not working actually. the hardware is not damaged. i am able to get the distance with an arduino. but the px4 firmware is not getting any information. i have checked it in nsh console. the cables are soldered like explained in your wiki: https://pixhawk.org/peripherals/rangefinder?s=lidar
i have attached the 5V to an external power supply.
I’m glad to see someone has replied to this post. I would be happy to provide logs and photographs of my flying environment if anyone has the time to look at them. Unfortunately I just disassembled my drone yesterday because I am going to overhaul the design of the frame.
@Joseph_Garret_Sulliv Have you tried unplugging your PX4Flow?
I noticed that PX4Flow does not work perfectly when a mocap system is flashing its LEDs, see this post PX4Flow disturbed by IR pulses of MOCAP