Discrete Event Control Previous: Introduction

Reverse Engineering and Inspection System

An integrated CAD/CAM/sensing system can be described as a closed loop (see Figure 1). We have developed a sensing system for reverse engineering which uses 2-d and 3-d vision to construct a CAD model of a part[4]. This sensing system interfaces with the University of Utah's _1modelling system which has a semi-automatic interface to automated manufacturing equipment.

Two-dimensional image processing routines are used to segment an image which contains an industrial part, and find closed contours in a pair of images (motion provided by robot arm). A stereo vision algorithm is used to obtain three dimensional data on the contours, and fitting and constraints are used to build a CAD model from the data. One of the resulting reverse-engineered parts is shown in Figure 2. The models used to manufacture these parts are shown in Figure 3. Note that the original model was not used in deriving the reverse engineered model, only sense data from the part itself.

Although the part is quite similar to the original, we would like to ``close the loop'' and use automated inspection with a CMM for a more accurate representation. To do this, we will inspect the original part, using our vision-derived model as the baseline. Doing so will require a robust control system as is described in the following sections.