Towards a Planning-Based Approach to the Automated Design of Chemical Processes

The design of chemical processes is a central problem in organic chemistry. A chemical process is a sequence of chemical reactions capable of producing a target compound from some starting stage molecules. The manual specification of such processes can be time-consuming and error-prone, due to the high number of reactions involved and their complex chemical conditions. To tackle this issue, we propose a planning-based approach and a framework for the automated design of chemical processes. Specifically, we argue that this problem can be reduced to a planning problem in Artificial Intelligence. To this end, we adapt the situation calculus and PDDL to the task of modeling molecules and capturing semantics of generic chemical reactions, and we conduct experimental study including empirical assessment of a PROLOG planner and two state-of-the-art planners on a set of benchmark problems.