search

opensim-org / opensim-moco

Solve optimal control problems for musculoskeletal models using OpenSim and direct collocation.
https://opensim.stanford.edu/moco
Apache License 2.0
56 stars 15 forks source link

Mesh refinement #27

Open chrisdembia opened 6 years ago

antoinefalisse commented 6 years ago

Do you already have plans for that? I am interested to know more about mesh refinement. I have read some stuff about how GPOPS deals with that (hp mesh refinement scheme, http://ieeexplore.ieee.org/document/7061603/ p51(16)). I think the number of collocation points is fixed using trapezoidal methods so that would be only a h mesh refinement scheme in muscollo I guess (to follow their terminology).

chrisdembia commented 6 years ago

Do you already have plans for that?

It's not an immediate priority, but I definitely want to do it.

I haven't read that paper; thanks for linking it. Yes, with the trapezoidal method, we'd only be able to do h mesh refinement.

I've implemented a very basic mesh refinement scheme (from Betts' book) on a toy problem, and it wasn't so bad. I think a simple scheme would work fine for most purposes.