POPCorn: Planning with Constrained Real Numerics

This dissertation outlines the work I have done since the beginning of my research degree. My research interest is in constrained resource planning, where I am particularly interested in the applications of operations research techniques in the task planning. Planning community has started to use operations research tools in their work in recent years. I aim to introduce new techniques to the task planning, which are widely practiced in operations research. The contribution of this paper is to present a generalisation of variables in the planning domain. We consider all types of predefined variables and the duration to a new type, which we call control parameters. We also describe the development of our new planner POPCORN (Partial-Order Planning with Constrained Real Numerics) that can reason with control parameters. We present an example of how existing task planning benchmark domains can be extended to develop enriched plans. We also provide an example to demonstrate the robustness and applicability of our approach.