Wed 5. Nov 2014, 17:25
AM-Robotics wrote:grab the cyan cylinder we observer a very high estimated external torque in some configurations (also in the static case)
please be careful to distinguish between
- external torque (topic /euroc_c2_interface_node/telemetry field measured.external_torque in msg MeasuredTelemetry, section 2.5.3.2 of the technical annex) and
- estimated external forces (service /euroc_c2_interface_node/get_estimated_external_forces response field external_force in srv GetEstimatedExternalForce, section 2.5.2.3 of the technical annex)
external torque is simply the difference between the measured joint torque and the expected joint torque. the expected joint torque is influenced via the /euroc_c2_interface_node/set_object_load service.
the
estimated external forces are calculated by using those measured
external torques and the current measured joint configuration (by using the pseudo inverse of the jacobian). this can not provide useful forces for arbitrary joint configurations.
if you observe high
external torque than this is caused either because of a collision (contact) or because a wrong "object load setting" via set_object_load.
if you observe high
estimated external forces this can be either because of a collision, wrong object load or because of the robots current joint configuration (for example: imagine the LWR in its 0-position)
AM-Robotics wrote:Is it possible that we call the service with the wrong parameters?
i'd suggest to move the LWR in a "reasonable" configuration and try to set different object loads - without the object actually grasped you will see how the LWR is deviated into "the opposite direction than what would be expected by that object load".
AM-Robotics wrote:Or does this service call not affect the external torque estimation?
this service call directly affects the
external torque calculation and by that also the
external force estimation.