Recursive Feasibility without Terminal Constraints via Parent-Child MPC Architecture

TL;DR

This paper introduces a hierarchical Parent-Child MPC framework that guarantees recursive feasibility and stability without terminal constraints, improving performance and scalability for nonlinear systems under uncertainty.

Contribution

The novel Parent-Child MPC architecture couples small and large horizon controllers to ensure recursive feasibility without conservative terminal constraints.

Findings

Enhanced computational efficiency over traditional MPC.

Reduced conservativeness in control design.

Scalable planning for nonlinear systems.

Abstract

This paper proposes a novel hierarchical model predictive control (MPC) framework, called the Parent-Child MPC architecture, to steer nonlinear systems under uncertainty towards a target set, balancing computational complexity and guaranteeing recursive feasibility and stability without relying on conservative terminal constraints in online decision-making. By coupling a small-horizon Child MPC layer with one or more large-horizon Parent MPC layers, the architecture ensures recursive feasibility and stability through adjustable stage-wise constraints derived from tube-based control. As is demonstrated in our case studies, compared to traditional MPC methods, the proposed Parent-Child MPC architecture enhances performance and computational efficiency, reduces conservativeness, and enables scalable planning for certain nonlinear systems.