Controller Startup Failures

[itaouil]

Ciao Gennaro :),

Thanks for open sourcing this work.

I am thinking to use your package as a way to make AlienGo climb stairs by integrating it with a planner.

I was trying to run the wbc in simulation by following the instructions in your README (through a system installation), but when I launch the package I obtain the following error:

But the controller_plugin.xml seems to be exported in the package.xml. Am I missing something?

Best, Ilyass

[graiola]

Hi @itaouil, it seems that a debian package was not correctly installed or sourced. Can you check if you have the debian wbc-setup/debs/bionic/advr/cartesian_interface-2.0.0-6d1e2e1-amd64-bionic.deb correctly installed? Also, do you have git-lfs?

[itaouil]

Hi,

Thanks for the prompt reply.

The debian package cartesian_interface-2.0.0-6d1e2e1-amd64-bionic.deb is present in the wbc-setup workspace.

And yes, I made sure to install git-lfs.

It actually installs the WBC debian packages, although it throws some warnings:

And similarly for the ADVR debian packages:

[graiola]

Can you roscd cartesian_interface?

[itaouil]

No.

I tried and there is no such package installed.

[graiola]

Can you confirm that the pkg is installed with dpkg -l | grep cartesian_interface? Also, could you try echo $ROS_PACKAGE_PATH ?

Thank you!

[itaouil]

The output of dpkg -l | grep cartesian_interface:

The output of $ROS_PACKAGE_PATH:

[graiola]

It looks like the xbot distro is not being sourced... try in a new terminal to run source /opt/xbot/setup.bash and then launch the wb_controller roslaunch wb_controller wb_controller_bringup.launch

[itaouil]

Yes. Sourcing the xbot setup fixed the issue :).

Thanks for you prompt replies and help.

[itaouil]

Hi @graiola,

In the latest push I think you might have forgot to include something while generation the debs:

[graiola]

Sorry, I should generate the debians in a docker container instead I am using my computer which sometimes has developments going on... I generated the new debians. Sorry for the inconvenience.

[itaouil]

No problem at all :).

Thank you!

[itaouil]

Problem seems to be still present :)

[graiola]

Give me few minutes...

[graiola]

Give it a try :)

[itaouil]

Error still present :)

[graiola]

Ok so I guess my environment is dirty... I would suggest you to use the old debian for the moment... I am going to work on it in the next hours

[graiola]

I just updated the debian again... hopefully it works this time :)

[itaouil]

Thanks for the push Gennaro.

While trying to install I receive this:

Before I try to downgrade the xbotinterface I want to make sure this is the right step? :)

[graiola]

I just tested the current configuration in the docker image, and it seems to work fine :)

but there is the new XBotInterface debian on git to download.

anyway our plan is to move the debians out of git in the next days to make the download process simpler

[itaouil]

Yes. The docker image works fine :).

There still seems to be some issues with the system install, but I will stick to using the docker image for the moment:

[graiola]

Hi @itaouil , we updated the way the debians are stored, now they are on dropbox. If you want, you can try to download the debians with install_dependencies.sh or get_debians.sh. Also, we removed git-lfs and purged the debians from the repo.

If you can give it a try it would be super helpful :)

Thank you!

[itaouil]

Hi,

Nice re-branding haha. I like the wolf part :).

The docker image works fine. Regarding this, where can I add a new world (under wb_gazebo_resources?) and a new aliengo model (I want to add a sensor plugin to it to have also some perception in the simulator along with the WBC)?

Regarding the system installation, I encountered some errors in the installation process:

1)

Even when a run the script with sudo

which I guess leads to this error:

2)

I tried to manually install the package by running sudo su and install the xbot env manually. but when I try that:

[graiola]

Hi @itaouil, thank you for the feedback :)

are you using install_dependencies.sh to install the debians or do you do it manually?

Regarding the gazebo model: you can add your new model to the $GAZEBO_MODEL_PATH, then it should show up when you launch gazebo. If you are using the docker, you can load your local ros workspace with the model in it with the option run_docker.sh --local your_ros_ws, but the container will not have the $GAZEBO_MODEL_PATH updated. You could try to modify line 15 in fun.cfg with:

-e GAZEBO_MODEL_PATH=$GAZEBO_MODEL_PATH:/opt/ros/melodic/share/wb_gazebo_resources/models/

so that the container gets your modified $GAZEBO_MODEL_PATH

if it works you can open a PR :)

Regarding the robot model, you can work in a similar way: just put everything in your local ros workspace and run the docker container with the local option. The way the robots are loaded is the following: There are an upload.launch and world.launch in wb_controller/launch that are called when the controller is spawned. These two launch files are responsible for loading the robot_description and robot_semantic_description on the ros param server which are the urdf and srdf files of the robot. So, technically you could imitate what I do with a robot urdf and srdf (for examples look at aliengo_description/robots) and load the new robot with roslaunch wb_controller wb_controller_bringup robot_name:=new_robot. Be careful that you will need also a set of params for the robot (look at wb_controller/params to get an idea).

I am going to move the robot params into each respective robot to simplify the loading :)

[itaouil]

Hi @graiola,

Thanks for the info.

Yes, I am using the script to install the dependencies.

The problem is similar to the one shown in the link. I tried to change to permissions before the install.sh script is run but with no luck so far.

Also, I noticed that you removed the XbotInterface-2.0.0 and now rely only on XbotInterface-2.1.0, but the xbot msgs is still at version 2.0.0, while xbot_env at version 1.0.0, could this be related to this error?

[itaouil]

Ok.

I managed to install it by --force-overwrite the installation of the packages, but then:

[graiola]

In the debs/advr folder, there is uninstall.sh, can you run it? and then run install.sh

[graiola]

ahaha why there are two robots? :):)

[itaouil]

In the debs/advr folder, there is uninstall.sh, can you run it? and then run install.sh

[graiola]

In the debs/advr folder, there is uninstall.sh, can you run it? and then run install.sh

it looks like a problem with _apt:

Give a look here

[itaouil]

In the debs/advr folder, there is uninstall.sh, can you run it? and then run install.sh

it looks like a problem with _apt:

Give a look here

Adding these two lines

sudo chown -Rv _apt:root $SCRIPTPATH/debs/$UBUNTU/advr/
sudo chmod -Rv 700 $SCRIPTPATH/debs/$UBUNTU/advr/

after the get_debians.sh call in the install.sh script fixes the installation error, but the two robot error spawn persists haha:

[graiola]

how are you running everything? In the docker, outside?

[itaouil]

No no the docker is fine, and works great.

All these errors are for the system installation only.

[graiola]

what if you spawn spot? robot_name:=spot

[graiola]

The docker image works fine. Regarding this, where can I add a new world (under wb_gazebo_resources?) and a new aliengo model (I want to add a sensor plugin to it to have also some perception in the simulator along with the WBC)?

I pushed the modification in run_docker.sh about the $GAZEBO_MODEL_PATH, feel free to try it

[itaouil]

In the debs/advr folder, there is uninstall.sh, can you run it? and then run install.sh

it looks like a problem with _apt: Give a look here

Adding these two lines

sudo chown -Rv _apt:root $SCRIPTPATH/debs/$UBUNTU/advr/
sudo chmod -Rv 700 $SCRIPTPATH/debs/$UBUNTU/advr/

after the get_debians.sh call in the install.sh script fixes the installation error, but the two robot error spawn persists haha:

This is what made the controller fail for me in the system installation process:

source ~/workspace/code/larvio_gpu/ros_wrapper/devel/setup.bashsource /opt/xbot/setup.bash

The sourcing command for xbot was added in the same line as the last one and not in a new one by the install_dependencies.sh script. Once I added an extra newline all worked fine for the system installation too.

[graiola]

That's weird because I explicitly append /opt/xbot/setup.bash in the bashrc... https://github.com/graiola/wolf-setup/blob/master/install_dependencies.sh#L62

Anyway, I am glad that the issue is fixed now.

Do you still have the problem with the two robots spawning?

[itaouil]

No everything works fine now :). No double robots party anymore.

[itaouil]

Hi :),

I changed the name of this issue so it accounts for all errors related to starting the controller.

In the last push the wolf_description_utils are not included so it cannot load the models:

Thought to let you know in case you missed it :)

P.S: If you might want to be notified about this sort of stuff via a different channel or by opening a new issue whenever the error is different. Let me know :)

[graiola]

I just added the debian :)

to speedup the communication we could use another channel too... suggestions?

[itaouil]

Yes. Now it works :).

Possible platforms could be:

• Telegram
• Slack
• Discord
• Mattermost

These are the ones that pop in my mind atm.