Robot position servoing using visual gap measurement

A typical robotic work space is rigidly structured. In practice, work pieces may be positioned less than perfectly, requiring special end-effectors and jigs. The introduction of a vision system allows work pieces to be more loosely positioned. In addition, a visual servoing system can make use of dynamic information gathered from the actual workpiece as it is picked up, positioned and worked on. This information can be used, for example, to bring two work pieces into alignment. In particular we describe a system capable of maintaining a steady width gap between a fixed and a movable workpiece. In this paper we use modular feature-based tracking. The application uses a hierarchical feature model. Features may be composed of several other sub-ordinate features. The main feature used in our application is one of a gap measurement type. This feature uses thresholding and moment calculations to determine the width of a gap presented in its region of interest.