Knowledge Representation and Efficient Image Processing for High-Level Skill Adquisition

Since it is important for service robots to be easily programmed, an interactive programming techniques like Programming by Demonstration (PbD) are obviously adequate programming methods. Image processing systems play an important role in PbD systems for acquiering the demonstration steps of the user. However, today's image processing systems are hardly capable of extracting relevant information from the user demonstration in real-time. In this paper, a multiresolutional approach will be presented to speed up the information acquiering process. Additionally, it will be shown how this information is used for creating an appropriate knowledge structure, which is convenient for representing high-level skills.