Application of Qualitative Reasoning to Robotic Soccer

摘要

In this paper we focus on qualitative reasoning aspects of knowledge representation for robotics and argue why they are appropriate for building robust control systems. We describe how to extract qualitative knowledge from a numerical world model, how to use this model for computing plans, and how to execute these plans. Robotic soccer is used as a real-world example domain. Problems might arise with regard to reactivity, unreliable sensor data might lead to stability issues and unintended behaviour. These drawbacks need to be treated. However, in general using a qualitative representation to do reasoning with offers several advantages such as robustness and efficiency.