OWL-enabled Assembly Planning for Robotic Agents

Assembly cells run by intelligent robotic agents promise highly flexible product customization without the cost implication product individualization has nowadays. One of the main questions an assembly robot has to answer is which sequence of manipulation actions it should perform to create an assembled product from scattered pieces available. We propose a novel approach to assembly planning that employs Description Logics (DL) to describe what an assembled product should look like, and to plan the next action according to faulty and missing assertions in the robot's beliefs about an ongoing assembly task. To this end we extend the KNOWROB knowledge base with representations and inference rules that enable robots to reason about incomplete assemblies. We show that our approach performs well for large batches of assembly pieces available, as well as for varying structural complexity of assembled products.