simtk.org

Hi,

I am looking at modeling lifting tasks with a full body model. In the simulations that I have performed so far, the model is moving in the vertical direction. I was wondering how I could get rid of this vertical motion as the feet do not actually move up and down and are just resting on the floor during the motion.

I tried adding a new "floor" body (connected to the ground) and welding the feet to it since the participants in the lifting tasks do not move their feet a lot. However, inverse kinematics did not work well with this extra constraint and all the coordinates were WAY off.

I then tried point constraint between a point on the ground and a point on the foot (2 point constraints with one for each foot). The location of the point on the ground body was based on the initial position of the subject, while the location of the point on the foot was just the COM of the toes. This seemed to work a bit better for inverse kinematics for the start of the motion. However, inverse kinematics quickly completely fall apart after that with unrealistic motion.

Since these two options did not work, I am not sure how I could restrict that vertical motion of the model during the lifting motion that is not actually happening?

Thank you

I had a similar problem with a model that I am working on as well. Have you looked at how ground-pelvis joint is defined? The model you are working with may only allow for rotation in the flexion extension direction. If the parent axis of the joint is placed anteriorly to the actual body of the pelvis (as it was in my original model), rotation about this axis will result in the lower segments being lifted off of the ground - mess around with the coordinate that allows for pelvic rotation to see if that is the case. For me, an easy solution was allow for translation in the ground-pelvis joint.

Hopefully this helps

Ryan