Closed traversaro closed 7 months ago
I guess that matlab-whole-body-simulator
should install a +mwbs.+Utils
directory, but I could not find it anywhere. Any idea on this @gabrielenava @nunoguedelha @VenusPasandi ?
After the fix https://github.com/robotology/whole-body-controllers/pull/145, now the simulation is failing with error:
MATLAB System block 'torqueControlBalancingWithSimu/Robot Visualizer/RobotVisualizer' error occurred when invoking 'setupImpl' method of 'mwbs.Visualizers.robotVisualizer.RobotVisualizer'. The error was thrown from '
'C:\Users\STraversaro\AppData\Local\mambaforge\envs\mwbs\Library\mex\+mwbs\+Visualizers\+robotVisualizer\RobotVisualizer.m' at line 53
'C:\Users\STraversaro\AppData\Local\mambaforge\envs\mwbs\Library\mex\+mwbs\+Visualizers\+robotVisualizer\RobotVisualizer.m' at line 26'.
Caused by:
Unable to resolve the name iDynTreeWrappers.loadReducedModel.
Component: Simulink | Category: Block error
However, this is an error as the whole-body-controllers
and/or matlab-whole-body-simulator
conda package does not depend on idyntree-matlab-bindings
, so it is actually a robotology-superbuild issue. Manually installing:
mamba install -c conda-forge -c robotology idyntree-matlab-bindings
permits to actually run the simulation.
However, running again the simulation fails with:
Build failed because the build file name(s) exceed the Windows limit of 260 characters. Build from a working directory with a shorter path, to allow build files to be created with shorter filenames. "C:\Users\STraversaro\AppData\Local\mambaforge\envs\mwbs\Library\mex\+wbc\examples\+floatingBaseBalancingTorqueControlWithSimulator\app\robots\iCubGazeboV2_5\slprj\_sfprj\torqueControlBalancingWithSimu\_self\sfun\src\c1_torqueControlBalancingWithSimu.cpp"
Component: MATLAB Function | Category: Build error
Build failed because the build file name(s) exceed the Windows limit of 260 characters. Build from a working directory with a shorter path, to allow build files to be created with shorter filenames. "C:\Users\STraversaro\AppData\Local\mambaforge\envs\mwbs\Library\mex\+wbc\examples\+floatingBaseBalancingTorqueControlWithSimulator\app\robots\iCubGazeboV2_5\slprj\_sfprj\torqueControlBalancingWithSimu\_self\sfun\src\c1_torqueControlBalancingWithSimu.cpp"
Component: Simulink | Category: Model error
Instead, this was due to my current working directory. Moving to C:\
worked fine.
I guess that
matlab-whole-body-simulator
should install a+mwbs.+Utils
directory, but I could not find it anywhere. Any idea on this @gabrielenava @nunoguedelha @VenusPasandi ?
Hi @traversaro, Utiles
is a class of matlab-whole-body-simulators
and it should be installed if you installed a version of matlab-whole-body-simulators
which is newer than the commit https://github.com/ami-iit/matlab-whole-body-simulator/commit/6078b4f2c1a874e2b48158998e311a9edf870bd5
I guess that
matlab-whole-body-simulator
should install a+mwbs.+Utils
directory, but I could not find it anywhere. Any idea on this @gabrielenava @nunoguedelha @VenusPasandi ?Hi @traversaro,
Utiles
is a class ofmatlab-whole-body-simulators
and it should be installed if you installed a version ofmatlab-whole-body-simulators
which is newer than the commit ami-iit/matlab-whole-body-simulator@6078b4f
I see. This means that we did a release of whole-body-controllers
(2.5.6
, done in https://github.com/robotology/whole-body-controllers/releases/tag/v2.5.6) without a compatible release of matlab-whole-body-simulator
. Probably we can close this issue as soon as we do a release of matlab-whole-body-simulator
.
I updated the issue to reflect the actual problem.
I think it would be nice to implement a github action that runs a test, at least for this Matlab-only controller, and give feedback on the controller status to understand in advance such problems. I don't know if it is possible to design a similar test also for Gazebo-dependent controllers.
There was already the idea of implementing CI in the repo long time ago: https://github.com/robotology/whole-body-controllers/issues/88
@VenusPasandi did something similar for an internal repo, and I think we can do something similar here. Changing the simulation time in https://github.com/robotology/whole-body-controllers/pull/146 from infinite to a finite number was actually meant to go in that direction.
the associated PR is closed with this comment:
https://github.com/robotology/whole-body-controllers/pull/145#issuecomment-1073827480
Hi @traversaro, I think this fix is not required. Actually, the function Rp2H is moved from the class State to Utiles in the commit https://github.com/ami-iit/matlab-whole-body-simulator/commit/6078b4f2c1a874e2b48158998e311a9edf870bd5.
I assume the issue can be closed, @traversaro feel free to reopen it if necessary!
I tried to run the
floatingBaseBalancingTorqueControlWithSimulator
by just runningfloatingBaseBalancingTorqueControlWithSimulator.torqueControlBalancingWithSimu
as documented in https://github.com/ami-iit/matlab-whole-body-simulator/tree/ac47383e3c04cd0cfd0ab11e2479b7991be9f491#eyes-checking-the-installation, and the simulation is failing with error:This are the version of the dependency that are installed: