TRANSFER OF CONTROL SKILL BY MACHINE LEARNING

Controlling complex dynamic systems requires skills that operators oft en cannot completely describe, but can demonstrate. This paper is conc erned with the problem of transferring human control skill to an autom atic controller The process of reconstructing a skill from an operator 's behavioural traces by means of machine learning (ML) techniques is called ''behavioural cloning''. The paper gives a review of ML techniq ues applied to behavioural cloning, a number of representative experim ents, and an assessment of the results. Some recent work is discussed, including a way of combining skill from several operators. (C) 1997 E lsevier Science Ltd. All rights reserved.