Robust adaptive MPC using control contraction metrics

We present a robust adaptive model predictive control (MPC) framework for nonlinear continuous-time systems with bounded parametric uncertainty and additive disturbance. utilize general contraction metrics (CCMs) to parameterize homothetic tube around nominal prediction that contains all uncertain trajectories. Furthermore, we incorporate adaptation using set-membership estimation. As result, the proposed MPC formulation is applicable large class of systems, reduces conservatism during online operation, guarantees constraint satisfaction convergence neighborhood desired setpoint. One main technical contributions derivation corresponding dynamics based on CCMs account state input dependent nature mismatch. optimize over parameter, which enables updates in MPC. Benefits are demonstrated numerical example involving planar quadrotor.