search

ToyotaResearchInstitute / idto

Inverse Dynamics Trajectory Optimization for Contact-Implicit Model Predictive Control
MIT License
72 stars 16 forks source link

Handle inequality constraints #29

Open kassasin opened 3 months ago

kassasin commented 3 months ago

Recently I readed this paper,I found there only handle equality contraint(dynamics),what can i do if i want to handle collision or joint limit?

vincekurtz commented 3 months ago

Great question, and thanks for your interest in our work!

The most straightforward way to handle inequality constraints would probably be a penalty method where the cost is augmented with a constraint violation term. The solver doesn't currently support this out-of-the-box, since we focused on the simplest possible case to optimize performance.

Updating the solver to support such penalties shouldn't be too difficult, but right now I don't have much time to work on it myself. If this is something you are interested in implementing we'd definitely welcome a contribution! We do have an (outdated, untested) prototype of joint and collision constraints in #18 that could be a good starting point.

kassasin commented 3 months ago

uderstand,I will take a try.