Efficient Qualitative Kinematics

A theory of qualitative kinematics is required for qualitative reasoning about the motion of objects in space . Such reasoning is important for many physical problem-solving tasks, such as design or robot plan ning. The practicality of earlier work has been limited by the complexity and non-compositionality of computing with configuration space constraints . This paper describes a much more efficient and compositional way for qualitative kinematic prediction . The method is based on a combination of earlier work on kinematic topology ( [4]), qualitative mechanism kinematics ( [3]) and subsumption detection . We show how each subtheory allows certain kinematic predictions to be made in a largely qualitative manner based on different object models. A complete qualitative kinematic analysis is obtained by combination of results obtained in the different models . This example shows how using several models of the same objects in parallel can yield significantly more powerful and efficient methodologies than were previously possible . Area: Qualitative Physics Subarea : Spatial Reasoning Submissions : Submitted only to IJCAI-91 and the 5th International Workshop on Qualitative Physics