Question on constraint forces (lambda)

[nhnhan92]

Hi everyone,
I have been trying to exact contact forces generated among objects colliding with each other. From what I understand that contact forces are equivalent to constraint forces (lambda) since I rely on the Lagrange multiplier method to infer contact response. However, I am still confused about the value I got from the GenericConstrainSolver interface. Is it true or not these values are solved to fulfill the constraint problem at the current time step (say, t_i) or final values fulfill the whole constrain problem from the beginning t_0 to the current time t_i. In other words, can we expect that these lambda values will be exact to the value measured by the force sensor or they are just incremental values (lambda_i) at time step 1_i (hence the final value should be a sum of all lambda_i).
Any clarification in codes, conceptual principles, etc. would be sincerely appreciated. Thank you all in advance.

[hugtalbot]

Hi @nhnhan92

indeed in the GenericConstraintSolver, you have two data: computeConstraintForces and constraintForces which are respectively a boolean to activate the computation of the constraint forces and the vector of constraint forces.

When you are running a simulation with collision using a DefaultContactManager with response="FrictionContactConstraint", a ConstraintSolver and a FreeMotionAnimation, the forces stored in the constraintForces data result from the constraint resolution. They correspond to the forces allowing to respect the given constraints at the current time step.

These forces will be used to compute a corrective motion of the free motion.
I hope this helps.
Best,

Hugo

[nhnhan92]

Thank you @hugtalbot for fast response.
From your explaination, It seems like if I want to compute the total contact force after a period of simulation time, I need to sum all constraint forces retrieved from GểnicConstrainSolver at each time step.
Correct me if I am wrong
Thank you for your help,
Nhan

[hugtalbot]

correct! 👍

[nhnhan92]

Thanks a lot