search

pal-robotics / reem_simulation

Packages for running REEM in the Gazebo simulator.
2 stars 9 forks source link

simulation fails to load #13

Open jon-weisz opened 9 years ago

jon-weisz commented 9 years ago

The simulator doesn't seem to load any more following the instructions from: http://wiki.ros.org/Robots/REEM/Tutorials/Launching%20a%20REEM%20Gazebo%20simulation under the current gazebo package (gazebo 1.9.6)

The torso spawns in the correct location, but winds up looking like this: http://imgur.com/g1ydMWd

No obvious errors show up in the gazebo log file. The controller reports the state of the torso joint as having NaN velocities.

@awesomebytes , this worked pretty recently. Any idea what might have changed? I've looked through the commit log, but I can't track it down.

awesomebytes commented 9 years ago

You catched it just in the middle of fixing it!

I'm done with reemc model (or almost) but not yet with reem. Tomorrow I should have both fixed!

Sorry (the problem is with the last version of gazebo and some values of the reem model), meanwhile I can only tell you to try to downgrade gazebo to 1.9.3 (or .1 or .2). On Oct 13, 2014 10:06 PM, "jon-weisz" notifications@github.com wrote:

The simulator doesn't seem to load any more following the instructions from: http://wiki.ros.org/Robots/REEM/Tutorials/Launching%20a%20REEM%20Gazebo%20simulation under the current gazebo package (gazebo 1.9.6)

The torso spawns in the correct location, but winds up looking like this: http://imgur.com/g1ydMWd

No obvious errors show up in the gazebo log file. The controller reports the state of the torso joint as having NaN velocities.

@awesomebytes https://github.com/awesomebytes , this worked pretty recently. Any idea what might have changed? I've looked through the commit log, but I can't track it down.

— Reply to this email directly or view it on GitHub https://github.com/pal-robotics/reem_simulation/issues/13.

jon-weisz commented 9 years ago

I've actually been beating my head against it all weekend. If you can tell me approximately what to look for, I can take a whack at fixing it this afternoon. Should I just look at the reemc model to see what changed?

awesomebytes commented 9 years ago

Yeah, take a look at public_sim branch there. It's inertias and something more.

But I'll try to upload something in the morning! Don't waste time :P On Oct 13, 2014 10:35 PM, "jon-weisz" notifications@github.com wrote:

I've actually been beating my head against it all weekend. If you can tell me approximately what to look for, I can take a whack at fixing it this afternoon. Should I just look at the reemc model to see what changed?

— Reply to this email directly or view it on GitHub https://github.com/pal-robotics/reem_simulation/issues/13#issuecomment-58950844 .

adolfo-rt commented 9 years ago

Adding on Sam's comments, the source of the problem is that the version of Gazebo has changed during the hydro cycle. Although it has only been patch fixes (1.9.x series), they have been significant enough to make the REEM simulation unstable. We had not realized this because we internally had Gazebo pinned down to a specific version in which the problem did not manifest itself.

As Sam mentions, he is currently making sure that the simulation can work with the most up-to-date packages released into hydro.

Thanks for your patience.

awesomebytes commented 9 years ago

As a temporary thing (I need further time today) you can use this rosinstall file: https://raw.githubusercontent.com/awesomebytes/pal-ros-pkg/master/reem-sim-hydro.rosinstall

It's working on my machine right now.

awesomebytes commented 9 years ago

This is going to stay on "stand by" until friday (but the version of the rosinstall works as far as I tested!) as I won't be able to work on it for two days. Sorry for the annoyance.

jon-weisz commented 9 years ago

@awesomebytes This seems to be a problem for everyone that provides gazebo simulations. Thanks for being so responsive!