multicopter hover thrust estimator use vehicle_thrust_setpoint (work in stabilized/manual mode) #19633
Conversation
dagar
changed the title
| @@ -166,10 +161,24 @@ void MulticopterHoverThrustEstimator::Run() | |||
|
|
|||
| _hover_thrust_ekf.predict(dt); | |||
|
|
|||
| vehicle_local_position_setpoint_s local_pos_sp; | |||
| vehicle_attitude_s vehicle_attitude{}; | |||
| _vehicle_attitude_sub.copy(&vehicle_attitude); | |||
There was a problem hiding this comment.
TODO: use all of these messages (vehicle_local_position, vehicle_attitude, vehicle_thrust_setpoint) on the same timestamp_sample?
There was a problem hiding this comment.
Given that we don't take the motor lag into account (the model assumes instant acceleration), I don't think that aligning the data perfectly is important.
|
Should we go even lower level and use the allocated_thrust from control_allocator_status? |
Il would improve the estimate on setups that are often saturating, yes. For the other ones, it shouldn't change anything. |
1 similar comment
Il would improve the estimate on setups that are often saturating, yes. For the other ones, it shouldn't change anything. |
|
Let's pick this up again. I've not seen the hover thrust estimator fail 👀
That's done 😃 |
…in stabilized mode)
mahimayoga
force-pushed
the
pr-mc_hover_thrust_stabilized
branch
from
0a90621 to
4d00981
Compare
Update the hover thrust estimator to use vehicle_thrust_setpoint (body frame) rather than vehicle_local_position_setpoint thrust (NED). This allows it to work in stabilized mode.
https://logs.px4.io/plot_app?log=6df3713e-f511-4e2e-854d-8fc7ab98ec65
TODO: