dynamicDrift computation -> seg fault

[andreadelprete]

I do not manage to understand why this code results in segmentation fault

from dynamic_graph.sot.hrp2_14 import robot
r = robot.Robot("hrp2");
r.dynamic.dynamicDrift.recompute(1);

Is there something wrong I am doing?

[olivier-stasse]

If you try to access a signal for which you did not specify the dependency, then the software is trying to access it without checking the dependency. This may create a seg fault.

[andreadelprete]

Thank you for your answer Olivier. What do you mean by "specify the dependency"? Do you mean "plug the signals on which dynamicDrift depends"? If that's the case, I think they are all automatically plugged inside the constructor of the python class Robot. Indeed if I call recompute on any other signal (e.g. com, inertia, Jcom) it doesn't seg fault.
Moreover, adding a print inside method computeTorqueDrift (line 1365 of file sot-dynamic.cpp) I found out that the seg fault is caused by the access to the vector Torques, which has (unexpectedly) size 0:

  newtonEulerSINTERN(iter);
  const unsigned int NB_JOINTS = jointPositionSIN.accessCopy().size();
  tauDrift.resize(NB_JOINTS);
  const vectorN& Torques = m_HDR->currentJointTorques();
  for( unsigned int i=0;i<NB_JOINTS; ++i ) tauDrift(i) = Torques(i);

[olivier-stasse]

Hi Andreas,
Happy new year !
Sorry for the delay in answering...
Regarding the snippet you gave, it might be due to the flag controlling the computation done in the sot-dynamic entity. The rational was to have a better control of what is computed in the control loop.
To have an example please look at the class AbstractHumanoidRobot and more specifically here

[andreadelprete]

Hi Olivier, happy new year to you too!
Don't worry for the delay, it's vacation time for everybody!
I'm sorry but I don't get your answer, maybe there's a misunderstanding here, so let me try to explain myself more clearly.

I'm trying to use the robot python class defined in dynamic_graph.sot.hrp2_14 to compute the dynamics of hrp2. I can compute Jacobians, forward kinematics, mass matrix, but when I try to compute the dynamicDrift I get a segmentation fault.
I see two possible reasons for this: either i) I'm using it wrong, or ii) there is a bug in the code of sot-dynamic. This is how I'm using it:

from dynamic_graph.sot.hrp2_14 import robot
r = robot.Robot("hrp2");
r.dynamic.dynamicDrift.recompute(1);

Is this the correct way of using it?