Continues rotating when takeoff at low speed about mav_dji_ros_interface HOT 3 CLOSED

Hello fx815,
Great to hear that you managed to make your drone work!

Did you send trajectories or a goal point via rqt topic publisher? If you haven't tested the latter, I recommend to test it (let's say sending 0, 0, 0.5 for x, y, z and leave rotation 0 0 0 1 ), as described in https://github.com/ethz-asl/mav_dji_ros_interface/wiki/First-Autonomous-Flight%3A-Hover-and-Step-Response.

If step response test works as expected (no rotating), then there may be some issues with the Polynomial trajectory publisher. Otherwise, some issues in the nMPC or other places.
Best,
Inkyu

from mav_dji_ros_interface.

Hi Inkyu,

I did try that before, and it worked. I suspect the rotating has something to do with the nmpc or disturbance observer. The drone I used was basically overloaded, it's quite heavy, plus the ground effect, I think it's quite hard for nmpc to control it takeoff as a low speed,say 1cm/s. When the drone is supposed to be off ground, the landing gear may still touch the ground, which may mess up the disturbance observer or nmpc somehow? The polynomial and poses topics are correct but the rollpitchyawratethrust topic given by nmpc has abnormal yaw rate values.
What do you think?

Best,
Feng

from mav_dji_ros_interface.

Hello Feng,
As you mentioned, this may be caused due to high payload. Rotating still happens with manual flight? If so, it seems the motors are saturated (out-of their controllable envelope). Maybe try to reduce the weight and try. According to M100 web page (https://www.dji.com/ch/matrice100/info#specs), the maximum payload is 1~1.2kg (3600-2355g).

from mav_dji_ros_interface.