Open traversaro opened 9 years ago
I now know what happened here, and probably this will also happen even with the modifications proposed by @HosameldinMohamed in https://github.com/robotology/whole-body-estimators/issues/17 . @HosameldinMohamed remember me to explain you.
If the external forces are properlty estimated from sensor readings, the residual coming from the computation of the dynamic step of the RNEA (including the estimated external forces) should be zero. Sometimes a warning appears saying that this residual is not zero, but this mismatch appear to be within the noise band of the estimation. My impression is that there is some sort of one-step mismatch between some of the estimated quantities and some of the quantities readed by the robot. Interestingly this happens only when readings skin contacts from the skinManager Example: