This is a planning grant in the research application program. Motivated by the need of modeling complex robotic environments, this project will explore new methods on object modeling/representation. Current modeling systems are limited to representing only the geometrical information of isolated objects, which is far less than sufficient. To be more effective and informative, a modeling system should also describe the changeable aspect of an object, i.e., the object states, the functionalities of an object, and the interrelations among objects. It seems, thus, reasonable to propose that modeling should be dynamic and sensor-related. The preliminary research and planning activities will focus on surveying the ground for, and identifying problems of dynamic modeling, which aims to make manipulating and reasoning about sensory information an integrated part of modeling, taking into account uncertainties. The objective is to gain sufficient insight about the subject in order to conduct more substantial research.
