A framework for behavioural cloning

Abstract This paper describes recent experiments in automatically constructing reactive agents. The method used is behavioural cloning, where the logged data from skilled, human operators are input to an induction program which outputs a control strategy for a complex control task. Initial studies were able to successfully construct such behavioural clones, but suffered from several drawbacks, namely, that the clones were brittle and difficult to understand. Current research is aimed at solving these problems by learning in a framework where there is a separation between an agent’s goals and its knowledge of how to achieve them.