Minimal offset between force plates and cameras
Posted: Sun Nov 08, 2015 3:09 am
Dear OpenSim community,
I would like to share an experience our team had with using RRA.
We were sure our lab was set up properly and the experimental data were correct which we used as an input into OpenSim. However, we received a systematic error after running RRA. The centre of mass (CoM) of the torso gets adjusted to account for errors between experimental GRF and joint angles. The CoM of the torso, however, got always shifted lateral to the left about 2-3cm, trial and participant independent.
As it always was the same side the CoM was shifted to and the error about the same size we started to wonder if there was a systematic error in our experimental data. After some research we found out that there was a minimal offset between our force plates and the camera system due to the calibration wand being missplaced for 1-2mm. This happened because the floor on top of the force plate was not 100% aligned with the force plate underneath it and when lying the wand we orientated the calibration wand to the floor and not the force plate (which was hidden under the floor).
This does cause not a big error on the first force plates, but as we have 4 force plates alligned to each other the error increased quite heavily at force plate 4. RRA is highly sensible to such errors and we therefore concluded that this calibration error may be the cuase for the CoM shift. We manually reorientated the force plates to the camera system which brought the CoM back to the middle of the torso. Interestingly, the muscle force predictions with static optimisation (without RRA) did not change drastically, the biggest change in magnitude could be found at the gluteus medius actuators. All muscle mainly responsible for the plane of progression did not change in magnitude or shape. But the CoM shift after RRA is not physiological correct and when using CMC it should be taken into account.
Therefore, it is important to be absolutely sure about the correct calibration wand placement for ecperimental data up to the milimeter. Also it would be better when having more than two force plates allingend behind each other to choose the global origin somewhere between these force plates to minimise the summation of calibration errors.
Best
Ursula
I would like to share an experience our team had with using RRA.
We were sure our lab was set up properly and the experimental data were correct which we used as an input into OpenSim. However, we received a systematic error after running RRA. The centre of mass (CoM) of the torso gets adjusted to account for errors between experimental GRF and joint angles. The CoM of the torso, however, got always shifted lateral to the left about 2-3cm, trial and participant independent.
As it always was the same side the CoM was shifted to and the error about the same size we started to wonder if there was a systematic error in our experimental data. After some research we found out that there was a minimal offset between our force plates and the camera system due to the calibration wand being missplaced for 1-2mm. This happened because the floor on top of the force plate was not 100% aligned with the force plate underneath it and when lying the wand we orientated the calibration wand to the floor and not the force plate (which was hidden under the floor).
This does cause not a big error on the first force plates, but as we have 4 force plates alligned to each other the error increased quite heavily at force plate 4. RRA is highly sensible to such errors and we therefore concluded that this calibration error may be the cuase for the CoM shift. We manually reorientated the force plates to the camera system which brought the CoM back to the middle of the torso. Interestingly, the muscle force predictions with static optimisation (without RRA) did not change drastically, the biggest change in magnitude could be found at the gluteus medius actuators. All muscle mainly responsible for the plane of progression did not change in magnitude or shape. But the CoM shift after RRA is not physiological correct and when using CMC it should be taken into account.
Therefore, it is important to be absolutely sure about the correct calibration wand placement for ecperimental data up to the milimeter. Also it would be better when having more than two force plates allingend behind each other to choose the global origin somewhere between these force plates to minimise the summation of calibration errors.
Best
Ursula