WPFM: The Workspace Prediction and Fast Matching System

An important area for the application of machine vision is to industrial assembly tasks, where the effective automation of such tasks is often dependent on the use of robots with sensing capabilities. The micro-world environment of a typical industrial assembly robot, although complex may also be highly constrained, with considerable information available a priori about objects and their disposition within a robot workcell. In order to maximise the efficiency with which a computer vision system can operate within this domain a priori information can be utilised to avoid extensive visual processing for constant scene features, or for objects in the scene for which there are accurate estimates of current position. The use of this information can then allow a reduction of the high computational cost of vision processing . This quantitative geometric information can be derived from a CAD based model of the robot workstation and from the known location of the camera system used to view the scene.