search

jrl-umi3218 / jrl-dynamics

Implementation of RNEA for robot control
ISC License
3 stars 3 forks source link

Error in NewtonEulerAlgorithm #4

Closed francois-keith closed 12 years ago

francois-keith commented 12 years ago

Hello

I think there is an issue with the commit on the master branch of jrl-dynamics (the commits adce9e0874 modifying the function NewtonEulerAlgorithm)

As a result, the following scripts crash as soon as the function "go" is called: sot-pg/python/walkingsimple.py, sot-dyninv/python/kinesimple.py, sot-dyninv/python/kinewalk.py . On the opposite, sot-dyninv/python/dyndebug.py seems not to be affected.

Regards

François Keith

aelkhour commented 12 years ago

Hello François,

Did your commit 1c54164 in sot-dyninv fix the issue? I must admit I am not familiar with neither sot-dyninv or sot-pg, although I'm responsible for the commit adce9e0 in jrl-dynamics.

Best, Antonio

francois-keith commented 12 years ago

Hello Antonio,

The commit in sot-dyninv is simply an update of the python scripts. The notation signal.sdes = ... was deprecated and needed to be replaced by signal.setReference(...) Unfortunately it does not solve the issue, but allowed me to see it, since they are my "reference scripts" for testing (although they are not in the release pack). What are the executable/scripts you use for testing?

François

aelkhour commented 12 years ago

Indeed the TestHumanoidDynamicRobot test in jrl-dynamics fails because spatial inertias are not computed by default when calling computeForwardKinematics, which leads to a bad computation of the center of mass with the default flags. As commit adce9e0 has no positive influence on current developments, I reverted it. Sorry for the inconvenience.

Antonio

francois-keith commented 12 years ago

Ok, the problem is not there anymore.

Thanks

François