Decentralised Collaborative Iterative Learning Control for MIMO Multi-Agent Systems

Abstract

Collaborative tracking control of multi-agent systems (MAS) involves two or more subsystems working together to perform a global objective, and is increasingly used within a diverse range of applications. However, existing, predominately centralised, control structures are sensitive to communication delays and data drop-out leading to inaccurate tracking. Moreover, comparatively little attention has been paid to the case of multiple input, multiple output (MIMO) linear agent systems. Iterative learning control (ILC) has been applied to increase tracking performance using past experience over repeated task attempts, but current ILC research assumes the ‘lifted’ system of each agent is full rank (i.e. each agent can achieve the task independently). This paper proposes a novel decentralised ILC framework, which can be applied to both full and non-full rank MIMO MAS. This framework provides powerful general conditions to design decentralised ILC laws. It is exemplified by application to derive three new decentralised ILC approaches: inverse, gradient and norm optimal ILC. Convergence and robustness analysis for the proposed framework are also given.