Zero-Order Robust Nonlinear Model Predictive Control with Ellipsoidal Uncertainty Sets

In this paper, we propose an efficient zero-order algorithm that can be used to compute approximate solution robust optimal control problems (OCP) and robustified nonconvex programs in general. particular, focus on OCPs make use of ellipsoidal uncertainty sets show that, with the proposed method, efficiently obtain suboptimal, but robustly feasible solutions. The main idea lies leveraging inexact sequential quadratic programming (SQP) which advantageous sparsity structure is enforced. obtained allows one eliminate variables associated propagation solve a reduced problem same dimensionality nominal OCP. drastically reduce computational complexity SQP iterations (e.g., case where exploiting interior-point method underlying (QPs), from O(N ? (nx6 + nu3)) nu3)). Moreover, standard embedded QP solvers for leveraged QP.