VISION

The focus of the project is set to the development of methods for visual analysis on a mobile robot platform, which has to operate in complex surroundings containing static as well as dynamic objects. The robot is equipped with sensors only, that find their counterpart among the senses of human being , i.e. mainly vision sensors.

The system has to detect the objects contained in the scene, distinguish between them and store them in a spatial and temporal local memory. Furthermore, it has to be able to predict the future movements of the dynamic objects (e.g. people). Based on these predictions the system is able to find its way through or around crowds of dynamic objects, avoiding collisions. To do so, an adequate strategy is necessary, which has to select between different possible actions.