Hi,
I took the MoBL_ARMS model and added a crank handle to the model. I attached a picture to the post. I limited the r_z coordinate to avoid mistakes during scaling and inverse dynamics. The IK results look correct but when I perform ID I get really high moments of more than 1000Nm. Does anybody know what could be the reason. Do I have to change anything with my crank handle joint? Without the included crank angle I do not get such obviously incorrect results.
Thank you very much.
MoBL_ARMS - Mistake in ID
- Pia Stefanek
- Posts: 48
- Joined: Tue Mar 13, 2018 12:21 pm
MoBL_ARMS - Mistake in ID
- Attachments
-
- Moments.PNG (23.38 KiB) Viewed 866 times
-
- MoBL_Arms.PNG (43.28 KiB) Viewed 866 times
-
- JointAngles.PNG (36.05 KiB) Viewed 866 times
Re: MoBL_ARMS - Mistake in ID
A couple of things come time mind:
1. Is it possible that the handle is imparting large moments on the hand that requires very large shoulder internal rotation and forearm pronation moments (e.g. to twist the handle) to oppose it? How is the handle modeled and which "r_z" coordinate was limited and how? What is the motion of the handle?
2. Often when we see moments in the 1000s there is a degree to radians conversion issue. If this model is run in 4.0, its possible that a coordinate that used to be labeled rotational was incorrect and is, in fact, coupled in the model. If you run IK and ID in 4.0 you should be fine. The discrepancy occurs when you run IK in 3.3 (coordinates incorrectly labeled rotational) and ID in 4.0.
1. Is it possible that the handle is imparting large moments on the hand that requires very large shoulder internal rotation and forearm pronation moments (e.g. to twist the handle) to oppose it? How is the handle modeled and which "r_z" coordinate was limited and how? What is the motion of the handle?
2. Often when we see moments in the 1000s there is a degree to radians conversion issue. If this model is run in 4.0, its possible that a coordinate that used to be labeled rotational was incorrect and is, in fact, coupled in the model. If you run IK and ID in 4.0 you should be fine. The discrepancy occurs when you run IK in 3.3 (coordinates incorrectly labeled rotational) and ID in 4.0.
- Pia Stefanek
- Posts: 48
- Joined: Tue Mar 13, 2018 12:21 pm
Re: MoBL_ARMS - Mistake in ID
Thanks for your reply Ajay,
I attached the model (already scaled) and the IK results to my post. Maybe you can have a look?
I tried to perform IK and ID without limitation of the r_z coordinate and the results were still comparable so this cannot be the reason for the high moments.
The person uses the handle to turn a crank.
I also attached the reaction file but I think everything is correct.
I performed the whole simulation in Opensim3.3 so I think there cannot occur degree/radian mistakes.
I hope you can help me.
Thanks,
Pia
I attached the model (already scaled) and the IK results to my post. Maybe you can have a look?
I tried to perform IK and ID without limitation of the r_z coordinate and the results were still comparable so this cannot be the reason for the high moments.
The person uses the handle to turn a crank.
I also attached the reaction file but I think everything is correct.
I performed the whole simulation in Opensim3.3 so I think there cannot occur degree/radian mistakes.
I hope you can help me.
Thanks,
Pia
- Attachments
-
- Model7scaled.osim
- (581.28 KiB) Downloaded 100 times
-
- IK_25_P3_25 3.mot
- (129.77 KiB) Downloaded 51 times
-
- H_P7_11_25W__REACTION (3).mot
- (20.61 KiB) Downloaded 55 times
- Pia Stefanek
- Posts: 48
- Joined: Tue Mar 13, 2018 12:21 pm
Re: MoBL_ARMS - Mistake in ID
Hi,
1. I found out that when I use multiplier functions in my coordinate set, the error decreases a lot. I discovered this effect by chance so maybe someone can help me to understand what the multiplier function exactly does. And why the function helps to decrease my ID error.
One example of the Multiplier function application:
<TransformAxis name="translation1">
<!--Names of the coordinates that serve as the independent variables of the transform function.-->
<coordinates></coordinates>
<!--Rotation or translation axis for the transform.-->
<axis>1 0 0</axis>
<!--Transform function of the generalized coordinates used to represent the amount of transformation along a specified axis.-->
<function>
<MultiplierFunction>
<function>
<Constant>
<value>0</value>
</Constant>
</function>
<scale>1</scale>
</MultiplierFunction>
</function>
</TransformAxis>
2. My ID results now look much better than before without the use of the multiplier function. But the unfiltered results still look weird (especially the elv_angle_moment and the shoulder_elv_moment). Is there still a mistake in there or are these normal results? I could not find any comparable ID results to check this by myself.
Thank you very much!
Regards
Pia
1. I found out that when I use multiplier functions in my coordinate set, the error decreases a lot. I discovered this effect by chance so maybe someone can help me to understand what the multiplier function exactly does. And why the function helps to decrease my ID error.
One example of the Multiplier function application:
<TransformAxis name="translation1">
<!--Names of the coordinates that serve as the independent variables of the transform function.-->
<coordinates></coordinates>
<!--Rotation or translation axis for the transform.-->
<axis>1 0 0</axis>
<!--Transform function of the generalized coordinates used to represent the amount of transformation along a specified axis.-->
<function>
<MultiplierFunction>
<function>
<Constant>
<value>0</value>
</Constant>
</function>
<scale>1</scale>
</MultiplierFunction>
</function>
</TransformAxis>
2. My ID results now look much better than before without the use of the multiplier function. But the unfiltered results still look weird (especially the elv_angle_moment and the shoulder_elv_moment). Is there still a mistake in there or are these normal results? I could not find any comparable ID results to check this by myself.
Thank you very much!
Regards
Pia
- Attachments
-
- ohnefilter.PNG (54.47 KiB) Viewed 786 times
-
- mitFilter.PNG (50.93 KiB) Viewed 786 times