Original Post
Hi. I''m working on a kinematics simulation of a 2D walking figure, using a Verlet-integrated particle system similar to the one described by Jakobsen in "Advanced Character Physics." I''m trying to implement angular constraints so the knees don''t bend backwards.
Jakobsen mentions using a distance constraint to keep the hip and foot away from each other, but after trying this technique I found that, though it''s stable, it doesn''t entirely address the problem (satisfying a distance constraint will give two possible solutions to a angle violation, one on either side of 180 degrees, so the knee can actually snap backwards given enough angular momentum).
Next I tried rotating the hip and foot points around the knee to change the angle when the constraint is violated, but this changes the momentum of the figure since the knee is never moved when solving a constraint. How can I change the angle trigonometrically while still keeping the center of mass consistent? Nothing I''ve found on the net addresses this issue. Please help.
--Duncan