Add SITL target for px4vision

[Jaeyoung-Lim]

This PR adds a gazebo SITL target for the Holybro Px4vision. More information can be found about the vehicle here

This PR adds a sitl airframe and updates PX4/sitl_gazebo submodule that includes PX4/PX4-SITL_gazebo-classic#440

Testing
The vehicle shown flying below

[dagar]

Any reason we can't share the actual airframe 4016? https://github.com/PX4/Firmware/blob/master/ROMFS/px4fmu_common/init.d/airframes/4016_holybro_px4vision

[TSC21]

Any reason we can't share the actual airframe 4016? https://github.com/PX4/Firmware/blob/master/ROMFS/px4fmu_common/init.d/airframes/4016_holybro_px4vision

That makes sense to me as well.

[Jaeyoung-Lim]

@dagar @TSC21 I still think we need a separate airframe config for sitl models since the mechanical properties / actuator configurations are all different. Wouldn't it be harder to maintain a single file that works on both?

IMHO, having the ekf2 masks and configurations (such as COM_OBS_AVOID) set properly would be already useful for people developing on the vehicle

[bys1123]

@Jaeyoung-Lim Then, Is that better if add "sitl" in file name, in case other developer edit wrong file.

[Jaeyoung-Lim]

@bys1123 It is already under a different directory. If you want to change the name, this would need to propagate also to all the models and make targets, which I find unnecessary

[bys1123]

Understood, thanks!

[TSC21]

@dagar @TSC21 I still think we need a separate airframe config for sitl models since the mechanical properties / actuator configurations are all different. Wouldn't it be harder to maintain a single file that works on both?

IMHO, having the ekf2 masks and configurations (such as COM_OBS_AVOID) set properly would be already useful for people developing on the vehicle

@Jaeyoung-Lim the usual procedure is to try to match the configuration on both sides, and if it's not possible, it needs to be checked what parameters make sense or not and what parameters need an adjustment. My recommendation is that you copy-paste the values of one config file to the posix one and test it. If it works right, then you can keep it.

[dagar]

@dagar @TSC21 I still think we need a separate airframe config for sitl models since the mechanical properties / actuator configurations are all different. Wouldn't it be harder to maintain a single file that works on both?

The goal would be to use that as a mechanism that forces the model to come into alignment with the real vehicle. If that's not realistic (or of interest) then maintaining both is fine.

[bys1123]

@Jaeyoung-Lim Any Update?

[bys1123]

@Jaeyoung-Lim Don't forget this, thanks~

[Jaeyoung-Lim]

I tried using the airframe configs of the real vehicle and maybe override the necessary changes on the sitl airframe if necessary. This will make the diff more explicit and manageable. However, the airframe configs seems to work out of the box for the px4vision.

@dagar @TSC21 @bys1123 Would this be a acceptable way to add the configs? I am not sure if this adds complexity to the configuration.

[TSC21]

Looks good. If one wants to add specifics to sim, can always add it in this config file.

[bys1123]

@Jaeyoung-Lim Thanks very much! How about keep

# param set COM_OBS_AVOID 1
# param set CP_DIST 6.0

but comment it, when user need it could easy to enable it?

[Jaeyoung-Lim]

@bys1123 The parameters are already enabled on the vehicle airframe configs here: https://github.com/PX4/Firmware/blob/master/ROMFS/px4fmu_common/init.d/airframes/4016_holybro_px4vision#L29 , https://github.com/PX4/Firmware/blob/master/ROMFS/px4fmu_common/init.d/airframes/4016_holybro_px4vision#L82

@TSC21 Thanks!

[bys1123]

Just re-think this, then good to go, Thanks!

[TSC21]

@Jaeyoung-Lim does it make sense to make use of this model for the Avoidance tests in CI? The iris with cameras is just and adapted model for the purpose and could make sense to use a model that is aimed for CV/obstacle avoidance instead.

[Jaeyoung-Lim]

@TSC21 I think that makes total sense. Would also get testing with real vehicles closer to the real vehicle

[TSC21]

@TSC21 I think that makes total sense. Would also get testing with real vehicles closer to the real vehicle

Is that something you can explore in this PR as well?

[Jaeyoung-Lim]

@TSC21 Would it be okay if I address it in a separate PR? I won't be able to work on it for the next few days, so I would rather close this quickly.

[TSC21]

@Jaeyoung-Lim yes that's fine. I can't merge because it is stating that the branch has conflicts.

[Jaeyoung-Lim]

@TSC21 Hmm that is strange. I resolved it with 557782f

[TSC21]

@TSC21 Hmm that is strange. I resolved it with 557782f

It must be Github doing its tricks again --'

[Jaeyoung-Lim]

@dagar Can we get this merged please?

[dagar]

@dagar Can we get this merged please?

deja vu 😏

[Jaeyoung-Lim]

@dagar Haha thanks!