CAD–based robotic vision

This paper presents an approach towards integrating three different technologies, namely computer–aided design, computer vision and robotics into one unified CAD–based robotic vision system for three–dimensional objects. The underlying principle of our approach to the design of the system lies in the concept of off–line programming or pre–computation. Three–dimensional CAD models constructed by the CAD system are used as the basis for recognising and locating objects. They are precomputed for their vision model during an off–line phase so that efficient and fast recognition of the object is possible during the run–time phase.