Evaluating PDDL for programming production cells

Abstract

A unique selling point for cyber-physical production system manufacturers becomes the easy with which machines and cells can be adapted to new products and production processes. Adaptations, however, are often done by domain experts without in-depth programming know-how. We investigate in this paper, the implications of using a planning-based approach for using a domain expert's knowledge to control the sequences of a robot and injection molding machine (IMM). We find that current engineering support is insufficient to address testing, understanding, and change impact assessment concerns during the evolution of a PDDL/HDDL domain specification.