Marcia K. O'malley

01-04-2002, 09:18 AM

Thanks to all for your responses. The initial posting asked if there

was a way to estimate the torques realized in each joint of the arm if

you know the force at the hand (for example, if you push on a solid

surface with a known force, how much torque is seen at each degree of

freedom of the arm?). The desired application was for force reflection

to a human via an arm exoskeleton. Normally you can calculate all the

joint torques if you know the jacobian of the exoskeleton and the force

to be displayed (from torque = transpose of the jacobian times your

force vector). However, this calculation gets cumbersome and can slow

down your sampling times as your degrees of freedom increase. I was

seeking a simpler method for estimating the torques at each joint based

on a known force vector. A summary of responses is included below.

-- M. O'Malley

>From Edward Morra (ed@orl-inc.com)

You can develop a simple model to calculate joint torques as a function

of hand load and joint angles, just work out a 2D free body diagram of

the arm and develop the static equilibrium equations from it. Examples

can be found in most biomechanics text books. If you have an equal

amount of equations as unknowns you can readily solve the problem. If

you have more unknowns than equations than you have to rely on some

optimization algorithm to get your answer. Most spreadsheet software and

certainly MatLab are effective tools in developing quasi-static

approximations in both cases.

>From Germano Gomes (G.T.Gomes@ncl.ac.uk)

If you want to use some freely available routines I recommend the

robotics toolbox for Matlab. It was developed by Peter Cork and is

available at:

http://www.cat.csiro.au/cmst/staff/pic/robot/

In particular, the function RNE (it uses a recursive Newton Euler

approach) should give you exactly what you want if you describe the

kinematics of the limb in terms of a D-H matrix (standard or modified).

Another package to consider is Spacelib developed by Legnani and

available in Matlab and C language from the ISB Technical Group on

Computer Simulation at:

http://www.isbweb.org/~tgcs/ Good luck

>From At Hof (a.l.hof@meg.rug.nl)

Some ideas might be got from my paper

A.L. Hof An explicit expression for the moment in multibody systems

J. Biomechanics 25: 1209-1211 (1992)

I hope this is what you need.

________________________________________________

Marcia K. O'Malley

Assistant Professor

Mechanical Engineering and Materials Science - MS 321

Rice University

6100 Main Street

Houston, TX 77005-1892

Tel: 713.348.3545

Fax: 713.348.5423

omalleym@rice.edu

http://www.ruf.rice.edu/~omalleym/

________________________________________________

---------------------------------------------------------------

To unsubscribe send SIGNOFF BIOMCH-L to LISTSERV@nic.surfnet.nl

For information and archives: http://isb.ri.ccf.org/biomch-l

---------------------------------------------------------------

was a way to estimate the torques realized in each joint of the arm if

you know the force at the hand (for example, if you push on a solid

surface with a known force, how much torque is seen at each degree of

freedom of the arm?). The desired application was for force reflection

to a human via an arm exoskeleton. Normally you can calculate all the

joint torques if you know the jacobian of the exoskeleton and the force

to be displayed (from torque = transpose of the jacobian times your

force vector). However, this calculation gets cumbersome and can slow

down your sampling times as your degrees of freedom increase. I was

seeking a simpler method for estimating the torques at each joint based

on a known force vector. A summary of responses is included below.

-- M. O'Malley

>From Edward Morra (ed@orl-inc.com)

You can develop a simple model to calculate joint torques as a function

of hand load and joint angles, just work out a 2D free body diagram of

the arm and develop the static equilibrium equations from it. Examples

can be found in most biomechanics text books. If you have an equal

amount of equations as unknowns you can readily solve the problem. If

you have more unknowns than equations than you have to rely on some

optimization algorithm to get your answer. Most spreadsheet software and

certainly MatLab are effective tools in developing quasi-static

approximations in both cases.

>From Germano Gomes (G.T.Gomes@ncl.ac.uk)

If you want to use some freely available routines I recommend the

robotics toolbox for Matlab. It was developed by Peter Cork and is

available at:

http://www.cat.csiro.au/cmst/staff/pic/robot/

In particular, the function RNE (it uses a recursive Newton Euler

approach) should give you exactly what you want if you describe the

kinematics of the limb in terms of a D-H matrix (standard or modified).

Another package to consider is Spacelib developed by Legnani and

available in Matlab and C language from the ISB Technical Group on

Computer Simulation at:

http://www.isbweb.org/~tgcs/ Good luck

>From At Hof (a.l.hof@meg.rug.nl)

Some ideas might be got from my paper

A.L. Hof An explicit expression for the moment in multibody systems

J. Biomechanics 25: 1209-1211 (1992)

I hope this is what you need.

________________________________________________

Marcia K. O'Malley

Assistant Professor

Mechanical Engineering and Materials Science - MS 321

Rice University

6100 Main Street

Houston, TX 77005-1892

Tel: 713.348.3545

Fax: 713.348.5423

omalleym@rice.edu

http://www.ruf.rice.edu/~omalleym/

________________________________________________

---------------------------------------------------------------

To unsubscribe send SIGNOFF BIOMCH-L to LISTSERV@nic.surfnet.nl

For information and archives: http://isb.ri.ccf.org/biomch-l

---------------------------------------------------------------