A feature-based inspection and machining system

Tom Kramer, Hui Huang, Elena R. Messina, Frederick M. Proctor, Harry A. ScottComputer-Aided 2001 · DOI

TL;DR

This paper introduces the Feature-Based Inspection and Control System (FBICS), an architecture for automating the machining and inspection of mechanical parts. FBICS leverages a standard feature-based description of part geometry to enable hierarchical process planning and execution, aiming for tighter integration and more flexible manufacturing processes.

What Problem Does This Paper Address?

Modern manufacturing, especially in job shops, faces challenges in integrating disparate systems and automating complex processes like machining and inspection. Traditional approaches often rely on manual programming and rigid, sequential process plans that are difficult for computers to interpret and adapt. This leads to inefficiencies, limited flexibility, and a lack of seamless data flow between design, planning, and execution. The paper addresses the need for a more integrated, intelligent, and flexible manufacturing system that can handle both machining and inspection tasks automatically, using standardized data representations.

Methods

FBICS employs a hierarchical control architecture based on the Real-Time Control System (RCS) model. Key aspects of its methodology include:

  • Feature-Based Representation: The system uses a standard feature-based description of part geometry, primarily adhering to STEP AP 224, which defines common manufacturing features like holes and pockets. This allows for a more abstract and intelligent representation of the part compared to raw geometric data.
  • Hierarchical Planning: FBICS implements a three-level hierarchical planning system (Cell, Workstation, Task) that automates process planning. This allows for breaking down complex tasks into manageable sub-tasks at different levels of abstraction.
  • Two-Stage Planning: To achieve flexibility and reusability, FBICS utilizes a two-stage planning approach. Stage-one plans are flexible, incorporating variables and alternatives, while stage-two plans are more concrete, derived from stage-one plans and tailored for execution.
  • Standardized Languages: The system utilizes standard languages for machine control and inspection: RS274 for machining and DMIS for inspection.
  • Open Architecture: FBICS is designed with an open architecture, facilitating the integration of different modules and subsystems.
  • Feedback Integration: A crucial aspect is the feedback loop from inspection to planning, allowing for adaptive machining based on measurement results.

Datasets

Not discussed in the paper. The paper focuses on the architecture and implementation of FBICS, using example parts like "simple_part," "TEAM," and "ANC101" for demonstration and testing, rather than formal datasets for training or benchmarking.

Key Insights and Ideas

  • Feature-Based Control is Key: Using standardized feature definitions (STEP AP 224) as the primary input for both machining and inspection significantly simplifies and automates the planning process.
  • Hierarchical Planning for Complexity: Breaking down the manufacturing process into Cell, Workstation, and Task levels mirrors how many job shops operate and allows for manageable automation of complex tasks.
  • Two-Stage Planning for Flexibility: The distinction between flexible stage-one plans and executable stage-two plans is vital for adapting to variations in manufacturing conditions and reusing plans effectively.
  • Inspection-Machining Integration: Tightly integrating inspection with machining, including using inspection feedback for adaptive control, is a significant step towards truly intelligent manufacturing.
  • STEP's Practical Utility: The paper demonstrates the practical application of STEP standards (specifically AP 224) in a real-world manufacturing control system, highlighting its value beyond just data exchange.

Key Results

The paper demonstrates the feasibility and functionality of FBICS through simulations and integrated tests. While specific quantitative results are not the primary focus, the paper highlights:

  • Successful Planning and Execution: FBICS successfully planned and simulated machining and inspection for moderately complex parts.
  • Efficiency in Planning: For a simple example part ("simple_part"), Cell-level planning took 15 seconds, and execution (including Workstation planning and simulated task execution) took 50 seconds on a Sun SPARCstation 20.
  • File Generation: For a more complex part like the TEAM part, the system generated hundreds of files, including NC code, DMIS code, and various plan and setup files, showcasing the system's comprehensive output.
  • Demonstrated Adaptability: The "finish_mill_adaptive" task demonstrates how inspection feedback can be used to adjust machining parameters for tighter tolerances.

Limitations and Future Work

  • User Interface: The authors acknowledge that user-friendly editors and interfaces for human participation have not yet been built.
  • Commercial System Integration: While the architecture is open, the paper notes that integrating with existing commercial systems can be challenging due to proprietary interfaces and data formats.
  • Full STEP AP 224 Implementation: While the core feature library of AP 224 is used, the boundary representation method is not.
  • Future Work: The paper suggests that the Task-level controller could be split into separate controllers for machining and inspection. Further integration with actual hardware and more complex parts is implied. The development of more robust user interfaces and the exploration of tightly-integrated versions of FBICS are also potential future directions.

Summary updated

BibTeX
@inproceedings{kramer2001feature,
  author = {Tom Kramer and Hui Huang and Elena R. Messina and Frederick M. Proctor and Harry A. Scott},
  title = {{A feature-based inspection and machining system}},
  booktitle = {Computer-Aided},
  year = {2001},
  doi = {10.1016/s0010-4485(01)00070-7},
}
Figure