PX4
QGroundControl
QGC
MAVSDK
MAVLink
Documentation
Docs
Support
Help

Delays in Raw Sensor Data(simulink-based HITL)

Flight Testing & Log Analysis
#1

Hey guys,

i am working on a simulink<->px4 communication for a simulink-based HITL at the moment. For testing purposes i send a sine signal as xaccel-data received by the pixhawk through the HIL_SENSOR-Message. In the log one can see that there are significant delays (about 0.04-0.08 sec)in the Raw-Accelerometer graph( and the other Sensors graphs as well) occurring periodically:

Screenshot (366)
Screenshot (366)854×512 37.7 KB

Am i right to suspect the simulink serial- interface to be responsible or could the problem somehow be on pixhawk input side(problem with buffer size/properties)?
If someone has any idea or advice i would be very thankful about.

Log: https://review.px4.io/plot_app?log=58288949-e5a8-475f-8e0b-e78639e49134

Kind regards,
Simon

#2

What interface are you using usb or serial ? i also working on a sinulink- px4 HITL ,mc Can be used normally ，but in fw ekf2 yaw has a erro, you can see uorb sensor publish rate (uorb top )

#3

i use usb connection like the jmavsim HITL does. To send data on this usb connection i use the ‘Serial Send’-Block in Simulink, which i suspect to cause the delays…

#4

this is my log https://logs.px4.io/plot_app?log=34b541a9-c54b-438b-b091-6795a229f522

#5

looking at your logs i am wondering, why it behaves quite stable, even if there are also significant delays in the sampled data, referring to the ‘Sampling Regularity of Sensor Data’ Plot

#6

Did you find lockstep in SITL，Used to eliminate delay in the sampled, looking at code src/board/px4/sitl/default.cmake

set(ENABLE_LOCKSTEP_SCHEDULER yes)

this is my topic PX4 + simulink HITL change yaw global position lost

#7

i use the hardware board with fmu-v5 build on it because i want to do ‘Hardware-in-the-loop’ . Lockstep works with sitl only, am i right?