I fixed the integration step size issue,
fixed a bug in the unilateral pulling force case in LinearPenaltyContactModel::computeForce
added a trot and a jump simulation,
for Euler simulation there are three options to compute the accelerations
pinocchio::aba()pinocchio::computeMinverse()
and solving M dv_ = tau_ - nle + J.T.f through cholesky decompositions
interesting enough, each one gives a different result .. the jump is only stable when using pinocchio::aba so I will have to retune it to make it more stable with the other methods.
The only reason I could think of is that solo12 legs are significantly lighter than the base, but I couldn't think of a way to find a ground truth to compare to.
The slipping point optimization and force prediction parts are commented in this pull request. I need to explore them a bit more.
I fixed the integration step size issue, fixed a bug in the unilateral pulling force case in
LinearPenaltyContactModel::computeForce
added a trot and a jump simulation, for Euler simulation there are three options to compute the accelerationspinocchio::aba()
pinocchio::computeMinverse()
and solvingM dv_ = tau_ - nle + J.T.f
through cholesky decompositions interesting enough, each one gives a different result .. the jump is only stable when usingpinocchio::aba
so I will have to retune it to make it more stable with the other methods.The only reason I could think of is that solo12 legs are significantly lighter than the base, but I couldn't think of a way to find a ground truth to compare to.
The slipping point optimization and force prediction parts are commented in this pull request. I need to explore them a bit more.