Optimal Sampled-Data Control of a Nonlinear System

TL;DR

This paper presents a novel stable-manifold approach for optimal sampled-data control of nonlinear systems, utilizing numerical techniques and trajectory interpolation to achieve effective control with long sampling periods.

Contribution

It adapts the stable-manifold method for sampled-data control, incorporating numerical computation of trajectories and derivatives, and introduces a shooting method for systematic trajectory generation.

Findings

Successfully applied to wheeled mobile robot tracking control

Effective with long sampling periods

Demonstrates the viability of the stable-manifold approach in sampled-data control

Abstract

Optimal sampled-data control of a nonlinear system is considered with the stable-manifold approach and extensive use of numerical techniques. The idea is to notice the Hamiltonian system associated with the considered optimal control problem and to compute trajectories on its stable manifold. Since the control input accompanied with those trajectories is proved to be optimal, the optimal control law can be obtained through interpolation. The stable-manifold approach was originally proposed for continuous-time optimal control and here it is adapted for sampled-data control based on the works of Navasca. In the case of sampled-data control, the approach requires the state transition of the controlled plant during one sampling period together with its derivatives with respect to the state and the input. Their computation is achieved by numerical techniques. Moreover, a shooting method is proposed for systematic generation of the trajectories and extension is considered for the intersample behavior to be taken into account. The proposed method is applied to tracking control of a wheeled mobile robot. It works successfully with a rather long sampling period.