A knowledge-based inspection workstation

We are building an inspection workstation development environment to use as a testbed for understanding what types of knowledge, e.g. data, algorithms and processes, can increase the productivity of inspection operations. Inspection can be more efficient through reducing the need for fixturing, integrating the generation of process plans and their execution within the controller, and reducing the errors or data losses that occur by translating the models to different formats. The initial configuration of inspection systems can be less costly through the use of open architectures that are constructed from components. Key elements of our work include in-situ feature-based planning, vision-driven part pose estimation and software methods to facilitate construction of manufacturing controllers. These provide a rich environment in which to study the categories of knowledge that are useful in intelligent control of inspection workstations. This paper describes our vision, approach and preliminary results