search

hjsuh94 / irs_mpc

iRS-LQR: iterative Randomized Smoothing LQR
MIT License
7 stars 0 forks source link

Thoughts on Contact Modeling #11

Open hjsuh94 opened 3 years ago

hjsuh94 commented 3 years ago

Broadly, we can consider three separate axis of how we should be modeling contact:

  1. Which order of dynamics should we consider?
    • Quasistatic Formulation (throw away the mass matrix)
    • Quasidynamic Formulation (throw away the momentum at next timestep)
    • Full second-order Formulation (consistent with Newton's law)
  2. How should we handle contact?
    • Penalty Method (application of stiff spring / dampers)
    • Event-driven integration of hybrid dynamics
    • Implicit time-stepping with LCP (Stewar t& Trinkle) / QP (made convex with Anitescu)
  3. What should the contact geometry be? (Closely related to how we should detect / quantify contact)
    • Point to point contact (sphere)
    • Point to surface contact / Surface to surface contact (Box / polytope geometries)
    • Hydroelastic contact
hjsuh94 commented 3 years ago

In my opinion, it is clear that hydroelastic will clearly stabilize the detected contact points, so we should be focusing more on penalty modeling and the order of the dynamics.

In addition, quasidynamic can just replace quasistatic because quasidynamic is more general, and quasistatic solutions might run into non-uniqueness problems. So we would end up with the following grid of options to try:

Contact / Dynamics Quasidynamic Second-order
Penalty ? Drake MBP
Implicit Pang's Simulator Stewart & Trinkle (Anitescu)

It is probably possible to make a quasidynamic simulation with the penalty method by simply throwing away the momentum at each timestep?