1: \begin{abstract}

2: 

3: Multibody simulation with frictional contact has been a challenging subject of

4: research for the past thirty years. Rigid-body assumptions are commonly used to

5: approximate the physics of contact, and together with Coulomb friction, lead to

6: challenging-to-solve nonlinear complementarity problems (NCP). On the other

7: hand, robot grippers often introduce significant compliance. Compliant contact,

8: combined with regularized friction, can be modeled entirely with ODEs, avoiding

9: NCP solves. Unfortunately, regularized friction introduces high-frequency stiff

10: dynamics and even implicit methods struggle with these systems, especially

11: during slip-stick transitions. To improve the performance of implicit integration for these systems we

12: introduce a Transition-Aware Line Search (TALS), which greatly improves the

13: convergence of the Newton-Raphson iterations performed by implicit integrators.

14: We find that TALS works best with semi-implicit integration, but that the

15: explicit treatment of normal compliance can be problematic. To address this, we

16: develop a Transition-Aware Modified Semi-Implicit (TAMSI) integrator that has

17: similar computational cost to semi-implicit methods but implicitly couples

18: compliant contact forces, leading to a more robust method. We evaluate the

19: robustness, accuracy and performance of TAMSI and demonstrate our approach

20: alongside relevant sim-to-real manipulation tasks.

21: \end{abstract}

22: