arXiv:2611.19660v1 Announce Type: new Abstract: Physics-based control is central to robot learning, but differentiating through implicit physics steps typically requires Hessian or Karush--Kuhn--Tucker (KKT) linear solves; we ask whether gradient steps suffice. Each simulation step minimizes an energy under many contact and collision constraints. Without constraints, we view control over $K$ steps as $K$-level optimization with strongly convex lower levels, recast as one min--max problem. Our Penalized Single-loop $K$-level method (PSK) uses only gradients and a projection, updates all levels in parallel, and reaches an $\epsilon$-stationary control in $O(\epsilon^{-3})$ iterations, with a $K$-independent exponent but a $K$-dependent constant. Contact-rich constraints, however, typically violate a constraint qualification (CQ). We replace each constraint by a smooth wall that acts on a band of constant width. Under mild assumptions that need no CQ, every exact solution of the banded lower levels is then feasible and banded Fritz--John, and PSK keeps the $O(\epsilon^{-3})$ complexity. Without a CQ that is uniform in the band width, however, a shrinking band loses the $K$-independent exponent: with width $\epsilon$ and fixed $K\geq2$, deterministic zero-respecting methods need $\Omega(\epsilon^{-(K+3)/4})$ calls on a hard instance whose control dimension grows with $1/\epsilon$. We prove our assumptions for smoothed kinematic and dynamic multi-robot navigation among segment obstacles. In navigation experiments, PSK, tuned per case, reaches every goal with sixteen levels and, at equal time, reaches a lower loss than tuned unrolling in 17 of 18 cases, while implicit differentiation is lower in 10.