In this thesis I have undertaken an investigation of plan generation for rapidly changing multi-agent environments. The thesis outlines the architecture for a planning system which integrates the traditional problem solving task with a process of 'dynamic goal creation' which underlies the planning system's ability to respond to the rapidly changing problem solving environment. The thesis identifies a distinction between characteristics of the problem solving task which result in goal-subgoal stability in the generated plan, and those which result in an unstable relationship. It is the latter which are addressed through dynamic goal creation for the purpose of achieving multiple goals.
This paper is not available online