Qualitative Motion Understanding

The vision system of an Autonomous Land Vehicle is required to handle complex dynamic scenes. Vehicle motion and individually moving objects in the field of view contribute to a continuously changing camera image. It is the purpose of motion understanding to find consistent threedimensional interpretations for these changes in the image sequence. We present a new approach to this problem, which departs from previous work by emphasizing a qualitative nature of reasoning and modeling and maintaining multiple interpretations of the scene at the same time. This approach offers advantages such as robustness and flexibility over "hard" numerical techniques which have been proposed in the motion understanding literature.