Achieving Autonomous Task Execution by Integrating a Combination of Ontologies and Semantic Languages Into a Human Factory Interface

Abstract

The term Industry 4.0 refers to the digitization and automation of factories. One research program addressing this topic, is the Factory of the Future project of the DLR. The aim of Factory of the Future is to develop a concept for a partly automated factory environment. In this context, the role of a robot in an industry can, on the one hand, be that of a robotic assistant within a secure human - robot collaboration. On the other hand, robots can collaborate with each other to execute tasks, fully autonomously. In addition, robots are able to be programmed flexibly and intuitively by non-experts. This enables a quick changeover of a robot's task and, subsequently, the production of smaller lot sizes, up to the individual product. This requires high agility from the robots, as they have to be able to execute a variety of different tasks while having only a short adapting period. This thesis presents an approach, which combines ontologies with domain independent semantic logic planning, making it possible to define tasks by their effects, thus, making robot programming more accessible to non-experts. Aside of that, the approach provides robots with the semantic-level flexibility they need to be able to autonomously perform a wide variety of functionalities. For this purpose, the modular capabilities of a robot are arranged in a certain sequence at runtime, so that their execution achieves the effect, defined in a certain task. To compute such a sequence, also referred to as plan, the modular capabilities, which are stored as skills in an OWL 2 ontology, are translated into the PDDL, along with the goal of a given task. The resulting PDDL domain - problem pair is subsequently used to parameterize a domain independent semantic logic planner, namely Fast Downward, which computes the necessary plan. After this sequence is translated into OWL 2, and processed further, a robot can execute it, and thus autonomously fulfill a given task it had never performed before.