Distributed Iterative Learning Control for Multi-Agent Systems With Nonidentical Trial Lengths

This paper is aimed at realizing the high-precision trajectory tracking task of a multi-agent system (MAS) subject to the nonidentical trial lengths. A distributed iterative learning control law is presented for each agent by leveraging the data of its nearest neighbors and the leader from the previous trials. A new virtual equivalent system approach is proposed such that a virtual equivalent MAS with the identical trial lengths can be established. Moreover, the convergence analysis of the actual MAS can be obtained by investigating that of the virtual equivalent MAS with the switching topologies. It is shown that with the increase of the iteration, the tracking error of each agent converges to zero if all agents can experience a full-learning trial frequently enough. A simulation example is also provided to illustrate our established distributed learning results.