In HIL mode, motor will be disarmed after 10s even if disable the auto disarmed function on the QGC

[kcx064]

In HIL mode, motor will be disarmed after 10s even if I disable the auto disarmed function on the QGC

[kcx064]

px4 version is 1.11.0

[dagar]

What are you using for HIL? Can you share the log with this happening? https://logs.px4.io/

[kcx064]

Thank you for your reply. I used Airsim v1.3.1 for HIL. The log is https://logs.px4.io/plot_app?log=28740df9-a979-40b0-82aa-bf7cf7cd6eb3

This issue maybe start with v1.11beta2

[kcx064]

My hardware is Cube Black and Durandal. They are both have the issue.

[Jaeyoung-Lim]

@kcx064 Can you try and build Airsim from source? I think this is most likely a Airsim issue since there was some issue with the v1.3.1 release of Airsim and was fixed afterwards

[kcx064]

@kcx064 Can you try and build Airsim from source? I think this is most likely a Airsim issue since there was some issue with the v1.3.1 release of Airsim and was fixed afterwards

Thank you! But PX4 v1.10.1 does not have the issue when I used Airsim v1.3.1

[Jaeyoung-Lim]

@kcx064 Yes, and that is why something had to be fixed on the Airsim side

[kcx064]

@kcx064 Yes, and that is why something had to be fixed on the Airsim side

It is to say PX4 v1.11 makes some modification for HITL mode which makes Airsim can not work properly in HITL mode, so Airsim need to be fixed?

[dagar]

Thank you for your reply. I used Airsim v1.3.1 for HIL. The log is https://logs.px4.io/plot_app?log=28740df9-a979-40b0-82aa-bf7cf7cd6eb3

This issue maybe start with v1.11beta2

"Disarmed by lockdown"

You're hitting this. https://github.com/PX4/Firmware/blob/af84f449c74b1b108bf23f6b8fdf63cb95c1b39a/src/modules/commander/Commander.cpp#L1681-L1692

In HIL lockdown is enabled for safety. Let me see about a quick fix.

@Jaeyoung-Lim we should talk about getting automated HIL tests in CI. One helpful step in that direction would be if we make it as trivial to launch HITL (gazebo or jmavsim) as SITL gazebo.

[Jaeyoung-Lim]

@Jaeyoung-Lim we should talk about getting automated HIL tests in CI. One helpful step in that direction would be if we make it as trivial to launch HITL (gazebo or jmavsim) as SITL gazebo.

Okay, makes sense. Do we want to hook this up to github actions?

[dagar]

The lockdown -> auto disarm was introduced in #14766.

[TSC21]

@Jaeyoung-Lim we should talk about getting automated HIL tests in CI. One helpful step in that direction would be if we make it as trivial to launch HITL (gazebo or jmavsim) as SITL gazebo.

@dagar @Jaeyoung-Lim,
@bperseghetti is already working on those. And he is facing the exact same problem as above. @bperseghetti do you mind syncing with @dagar on a fix for this? Thanks!

[dagar]

@Jaeyoung-Lim we should talk about getting automated HIL tests in CI. One helpful step in that direction would be if we make it as trivial to launch HITL (gazebo or jmavsim) as SITL gazebo.

Okay, makes sense. Do we want to hook this up to github actions?

Let's get the pieces in place to do it as a developer first, then figure out how to make it live in github actions or Jenkins (attaching to hardware is a "detail").

1. Is sitl_gazebo HIL good enough to start using regularly? How can we make it trivial to launch directly and specify the serial port? What about included airframes that correspond to sitl_gazebo models? I don't think they're in the NuttX ROMFS.
2. Scripting the process.
   • flashing, setting the right airfarme, connecting
   • launch mission/mavsdk test and monitor
   • transfer ulg over mavlink

We can keep discussing outside of this issue.

[dagar]

@kcx064 can you try #15690?

[kcx064]

@kcx064 can you try #15690?

It's my pleasure!

[kcx064]

@kcx064 can you try #15690?

Thank you! I have try your branch, it work properly in hil mode!

[bperseghetti]

@dagar @TSC21 @Jaeyoung-Lim I had noticed this issue with HITL using gazebo last Monday. I think there is a lot of ways we can automate HITL testing. I had "ugly fixed" it by just commenting out the autodisarm and was looking through the rest of the HITL pipeline to see more about avoiding edge cases and how well it can "represent" the actual system before issuing an actual fix for that specific bug. I pulled and tested the branch for the bug fix using HITL/gazebo and it works as well for FMUv5_default using HITL_quad (1001) with modified iris.sdf model in gazebo.