Approximate Tracking Controllability of Systems with Quadratic Nonlinearities

TL;DR

This paper investigates the approximate tracking controllability of linear and quadratic systems, showing limitations for linear systems and establishing weak controllability results for nonlinear systems with applications to motion planning.

Contribution

It demonstrates the impossibility of full state tracking in linear systems and proves weak approximate tracking controllability for nonlinear systems with quadratic nonlinearities.

Findings

Linear systems cannot achieve approximate tracking controllability except in trivial cases.

Weak approximate tracking controllability is achievable for nonlinear systems with quadratic terms.

Applications include motion planning and handling systems with finite-time singularities.

Abstract

Given a finite-dimensional time continuous control system and $\varepsilon>0$, we address the question of the existence of controls that maintain the corresponding state trajectories in the $\varepsilon$-neighborhood of any prescribed path in the state space. We investigate this property, called approximate tracking controllability, for linear and quadratic time invariant systems. Concerning linear systems, our answers are negative: by developing a systematic approach, we demonstrate that approximate tracking controllability of the full state is impossible even in a certain weak sense, except for the trivial situation where the control space is isomorphic to the state space. Motivated by these negative findings for linear systems, we focus on nonlinear dynamics. In particular, we prove weak approximate tracking controllability on any time horizon for a general class of systems with arbitrary linear part and quadratic nonlinear terms. The considered weak notion of approximate tracking controllability involves the relaxation metric. We underline the relevance of this weak setting by developing applications to coupled systems (including motion planning problems) and by remarking obstructions that would arise for natural stronger norms. The exposed framework yields global results even if the uncontrolled dynamics might exhibit singularities in finite time.