Inverse kinematics results

Provide easy-to-use, extensible software for modeling, simulating, controlling, and analyzing the neuromusculoskeletal system.
User avatar
Marta Meneses
Posts: 26
Joined: Thu Mar 03, 2011 2:49 am

Inverse kinematics results

Post by Marta Meneses » Mon Sep 12, 2011 7:52 am

Hi everyone,

In output file from Inverse Kinematics file, the coordinates "*_tx" ,"*_ty" "*_tz" are expressed in which unit??

Thanks,

Marta

User avatar
Rudolf Klemetti
Posts: 48
Joined: Thu Jul 15, 2010 3:09 pm

Re: Inverse kinematics results

Post by Rudolf Klemetti » Mon Sep 12, 2011 8:38 am

Meters.

User avatar
Marta Meneses
Posts: 26
Joined: Thu Mar 03, 2011 2:49 am

Re: Inverse kinematics results

Post by Marta Meneses » Tue Sep 13, 2011 6:30 am

Thanks!

And what about the other coordinates?For instance, I've a coordinated define "shoulder_rot_l" whose bound values are [-90,90], but in the motion file I've obtained from IK its values reaches -430...
And I don't understand what's happening...

User avatar
Rudolf Klemetti
Posts: 48
Joined: Thu Jul 15, 2010 3:09 pm

Re: Inverse kinematics results

Post by Rudolf Klemetti » Tue Sep 13, 2011 6:50 am

Angles are in degrees. Since one revolution is 360 degrees, 90 + 360 = 430, so your angle is actually 90 degrees but its value contains one extra revolution - for some reason I don't know. Mathematically, that shouldn't be any problem as far as changes of angles are correct. I don't know if OpenSim realizes that equality of angles though.

User avatar
Marta Meneses
Posts: 26
Joined: Thu Mar 03, 2011 2:49 am

Re: Inverse kinematics results

Post by Marta Meneses » Tue Sep 13, 2011 9:31 am

That explanation works in this case...but I have other situation where the same happens, like, the angle is supposed to be 130º, but what appear in the motion file is around 535

User avatar
Ayman Habib
Posts: 2244
Joined: Fri Apr 01, 2005 12:24 pm

Re: Inverse kinematics results

Post by Ayman Habib » Tue Sep 13, 2011 11:43 am

Hi Marta,

What version of OpenSim are you using? Some earlier versions of OpenSim allowed users to specify an Optimizer and some optimizers didn't respect the bounds on coordinates but the latest versions do build coordinate bounds into the IK problem so it's unlikely the solution would violate them. I'm not sure where you specify coordinate bounds in your case (if from GUI or in the XML file). If from XML please keep in mind that angles are specified in radians (not degrees).

Hope this helps,
-Ayman

User avatar
Marta Meneses
Posts: 26
Joined: Thu Mar 03, 2011 2:49 am

Re: Inverse kinematics results

Post by Marta Meneses » Wed Sep 14, 2011 8:29 am

I'm using OpenSim 2.2.1, and specifying the bound ins XML file, but I've had in mind that the values are in radians! As for the Optimizer, which ones did no respect the bounds?!

One more question, can you tell me what is the sing convention for moments of force?

Thanks!

User avatar
Michael Sherman
Posts: 807
Joined: Fri Apr 01, 2005 6:05 pm

Re: Inverse kinematics results

Post by Michael Sherman » Wed Sep 14, 2011 10:04 am

martam wrote: One more question, can you tell me what is the sign convention for moments of force?
We use the right-hand rule.

Sherm

User avatar
Ayman Habib
Posts: 2244
Joined: Fri Apr 01, 2005 12:24 pm

Re: Inverse kinematics results

Post by Ayman Habib » Wed Sep 14, 2011 2:04 pm

Hi Marta,
martam wrote:I'm using OpenSim 2.2.1, and specifying the bound ins XML file, but I've had in mind that the values are in radians! As for the Optimizer, which ones did no respect the bounds?!

One more question, can you tell me what is the sing convention for moments of force?

Thanks!
If you specify "Jacobian" as an Optimizer Algorithm in 2.2.1 and earlier versions then it runs faster but doesn't respect the bounds, if however you didn't specify an Optimizer or used IPOPT then it does respect bounds. Later versions starting at 2.3.x do not give a choice of optimizer, instead it finds the best one for the problem internally and builds the bounds into the IK problem so they are always enforced, please give that a try. Keep in mind that the optimizers take into account the "clamped" flag of the coordinates.

As to the signs for moments and forces (e.g. ID output) you basically need to look at how the coordinate is defined, +ve forces/moments act along the direction of coordinate increase and -ve forces/moments act in the opposite direction.

Hope this helps,
-Ayman

User avatar
Marta Meneses
Posts: 26
Joined: Thu Mar 03, 2011 2:49 am

Re: Inverse kinematics results

Post by Marta Meneses » Thu Sep 15, 2011 6:31 am

Actually I did specified "Jacobian" optimizer....I'll try to run it all over again with the "ipopt" optimizer, and hope it works right now!So, you said Jacobian was the faster optimizer, and it took an hour to run the analysis, how long would it take with the "ipopt"??

Thanks a lot for you help!

POST REPLY