[DO_NOT_MERGE_YET] rotors_hil_interface: take advantage of MAVROS hil plugin

[TSC21]

With this PR we take (almost) full advantage of MAVROS hil plugin. There, the unit conversions are applied, besides the frame transform to NED. I also applied the uncrustify style for C++. Let me know if you want me to revert this step.

[ethzasl-jenkins]

Can one of the admins verify this patch?

[TSC21]

@hitimo @pvechersky @helenol can you please take a look? Thanks

[TSC21]

Related #356, #367, #380, #402.

[TSC21]

My only question is here: on MAVROS, we apply ENU to NED transform (https://github.com/mavlink/mavros/blob/master/mavros/src/lib/ftf_frame_conversions.cpp#L30) and then base_link to body/aircraft frame (https://github.com/mavlink/mavros/blob/master/mavros/src/lib/ftf_frame_conversions.cpp#L37), while you just seem to apply this last one. Which frame of reference are we using in rotors_gazebo?

[pvechersky]

I believe Rotors uses the standard ROS reference frame of NWU, hence the rotation by 180 degrees around the x axis to convert it to NED in the HIL interface node in this case.

[TSC21]

@hitimo @pvechersky @helenol can you give me your thoughts on this?

[pvechersky]

Conceptually, the changes are good. I think as much of MAVROS's built-in functionality should be leveraged by our HIL module, and using the hil plugin is a good step in that direction.

I just have a couple of higher-level comments that the other reviewers might follow-up on:

1. In rotors_hil.rosinstall you change the MAVROS uri to the mavlink repo instead of the ethz-asl one. The switch to ethz-asl repo for MAVROS was done after my time of working on this HIL interface so I'm not sure if the Rotors admins would prefer to stick to the ethz-asl repo or not. But if the ethz-asl repo is preferred then perhaps relevant hil plugin changes in MAVROS can be merged into that repo?

2. Code formatting. Rotors used Google coding standards I believe. See other C++ code in the project for reference.

Thank for integrating these improvements!

[ethzasl-jenkins]

Can one of the admins verify this patch?

[helenol]

ok to test

[protobits]

Hi, I would like to use HIL and I'm blocked by #367 which I understand would be solved by this. What is needed to finish this PR?

[protobits]

Hi, anyone still looking into this? I tried this PR and everything mostly works. I have only two issues:

1. mavros does not send hil_controls by default, it now send hil_actuator_controls so the first needs to be enabled somewhere or this interface adapted to parse hil_actuator_controls
2. I tried both sensor and state level HIL and I'm not getting response from RC when raising throttle in manual mode. The actuator controls message even oscilates between all zeros and 0.5 for the first four controls.

[TSC21]

@v01d I got too much on my back to finish this, as I am also supporting and maintaining sitl_gazebo PX$ package. I will be happy to help you in debugging your problems though.

[protobits]

Cool thanks.
Well, the problem right now is that I cannot get the pixhawk to send anything but zeros on HIL actuator controls. I can arm and the actuator targets are set when changing throttle stick, but something inside pixhawk is not right. I believe I'm sending everything for manual control (IMU, MAG, pressure) so it should work. How can I debug this? I don't receive any strange message from pixhawk either.

[protobits]

Another question, is the mavlink interface plugin required? I'm not really sure what it is for, since the hil interface seems to do all the talking to mavros.

[Jaeyoung-Lim]

@TSC21 How should we move forward with this?

[Jaeyoung-Lim]

test this please