Closed francesco-romano closed 7 years ago
This does not happen on iCubGenova02
It happend for iCubHeidelberg01 today. I think before I did not notice it because I used to launch it from the manager, today I launched it from terminal and on exit it gave Segmentation fault.
Does it happen every time?
I'm not sure, I have to check. I launched WBD3 only once from the terminal today. Can try that again.
:+1:
Yes it happens everytime, just tried a couple of runs and it always exits with Segfault.
Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0x7fff63fff700 (LWP 19234)]
0x000000000042c2a9 in wholeBodyDynamicsThread::closePort(yarp::os::Contactable*) ()
(gdb) bt
#0 0x000000000042c2a9 in wholeBodyDynamicsThread::closePort(yarp::os::Contactable*) ()
#1 0x000000000042c32f in wholeBodyDynamicsThread::closeOdometry() ()
#2 0x0000000000431642 in wholeBodyDynamicsThread::threadRelease() ()
#3 0x00007ffff7b67a95 in theExecutiveBranch(void*) () from /usr/local/src/robot/yarp/build-x86_64/lib/libYARP_OS.so.1
#4 0x00007ffff599cb16 in ACE_OS_Thread_Adapter::invoke() () from /usr/lib/libACE-6.0.3.so
#5 0x00007ffff5499182 in start_thread (arg=0x7fff63fff700) at pthread_create.c:312
#6 0x00007ffff5d2447d in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:111
(gdb)
I am unable to reproduce this issue.. : [
The configuration files for yarpWholeBodyInterface
and wholeBodyDynamicsTree
are the one in the repositories?
This https://github.com/robotology/codyco-modules/issues/163 should fix it.
@hu-yue can you test this?
Yes!
@francesco-romano @traversaro the fix does fix the segfault issue, however, I had problems with torqueBalancing
on right leg. The left leg fell despite all the parameters are correctly set. In fact, going back to the master branch of codyco-modules
the leg did not fall.
So probably this is an issue with the base. Can you paste here the configuration of the balancing and of the yarpWhole etc .ini file? (of this second file, you can avoid pasting the joint list.. I'm curious about the configuration of the base
balancing
name torqueBalancing
robot icub
period 10
modulePeriod 0.01
wbi_config_file yarpWholeBodyInterface.ini
wbi_joint_list ROBOT_TORQUE_CONTROL_JOINTS
constraint_links ("r_sole")
comKp (40 47 40)
comKd (0 0 0)
comKi (0 0 0)
kw 1
kImp (100 150 100 30 30 30 30 40 40 100 70 65 30 20 20)
tsat ( 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24)
torqueGains torqueGainsRightStanding.ini
yarpWholeBodyInterface first rows
robot icub
urdf model.urdf
getLimitsFromControlBoard
[WBI_STATE_OPTIONS]
localWorldReferenceFrame r_sole
estimateBaseState
Don't know if needed, wholeBodyDynamicsTree first rows
name wholeBodyDynamicsTree
period 10
robot icub
cutoff 3.0
#This file is referenced here, but it
# should be found in a robot specific directory
wbi_conf_file yarpWholeBodyInterface.ini
Btw, these are local files, they are not the same as in the repository ever since we changed the torqueBalancing on single leg from left to right leg.
Thanks @hu-yue Let's wait for @traversaro thoughts. If needed we can also Skype/Hangouts
There is too much cruft that accumulated related to base link
/calibration stuff.
Doing a bit of open loop cleaning is probably the most efficient way of understanding what is going on, I will come back to you to test again after a first round of cleaning.
In the day I am busy with TU/e related tasks, and this weekend I can't work on it. Any fix is delayed until one of the next week evenings.
@hu-yue are you ok to switch to master and keep the old behavior for the moment ?
@traversaro totally fine, it's not urgent for us
Won't fix, wholeBodyDynamicsTree will be deprecated soon as descriebed in https://github.com/robotology/codyco-modules/issues/223 .