A fully automated production system for higher throughput and improved quality

Machinery | Alignment | Blob Analysis | Calibration | Classification | Filtering | Inspection | Matching | Measuring | Morphology | Object Detection | Subpixel

MSTVision GmbH, a MVTec Certified Integration Partner, is an independent service provider with focus on the area of industrial image processing and automation. Based on their long year experience, they develop algorithms and scripts using the HALCON software library from MVTec. Typical projects are:

  • Development of state of the art algorithms, systems or submodules meeting the needs of their customers
  • Implementation of the required image processing using HALCON and/or pre-processing using the VisualApplets developer environment
  • Carrying out feasibility tests and concept studies including test set-ups adapted to customer requirements
  • Workshops at the customer’s site or at MSTVision’s offices (e.g. development workshops, trainings regarding HALCON, VisualApplets, line scan cameras and more)
  • Support of all phases of procurement projects
  • Tasks related to quality assurance and process optimization

From microscope-assisted manual assembly to fully automated process

HUBER+SUHNER Cube Optics AG located in Mainz, Germany, is developing, manufacturing and marketing highly miniaturized fiber optic components, modules and systems for use in data and telecommunication products. Up to now, most products are assembled in manual processes. To meet the requirements of future mass markets, fully automated processes are necessary.

Therefore, the overall goal of the project was to increase the throughput with minimized production rejects and improved quality control.

As a solution, an automated assembly machine has been introduced to the manufacturing process. At multiple inspection stations, 12 Megapixel cameras, high magnification telecentric lenses with field of view of a few millimeters in combination with various illuminations are installed. To ensure precise positioning, communication to motorized stages for linear and angular movements with µm precision, as well as process equipment (motorized grippers, dispensers for adhesive, etc.) and a SCARA robot for pick and place operations of material has been established.

HALCON is integrated into many different process steps. The software is used to identify parts (mostly made of glass), detect their positions and measure characteristic features with µm precision. Elements with unknown or varying shape (e.g. liquid adhesive from a dispenser) are tracked and identified within complex images.

The fully automated system leaded to a high throughput and, due to full process control, also to higher quality and higher yield. Only the accurate image processing provided the various parameters for controlling the µm-precise movements of the various actuators during the complex assembly process.

Challenges

During the complex integration process, some image processing challenges had to be overcome:

  • Difficult image content due to:
    • Flexible fibers
    • Shiny metal parts with changing texture
    • Parts made of glass (transparent and semitransparent with changing surface properties)
    • Adhesive with changing shape and changing properties over process time
    • Contaminations like dust and particles
    • Moving parts, moving background, moving manipulators
  • Changes in characteristics/appearance of parts in process due to fabrication tolerances in the incoming material
  • An image processing library, which meets the needs of the current as well as future projects, was required
  • Flexibility to adapt to customer-specific designs with as little effort as possible was required

Why HALCON

Before the project started, another machine vision software was used. However, tough limitations occurred when it came to complex image content. Interfacing to an industrial automation platform like C#/.NET and EtherCAT was not as easy anymore. To overcome these obstacles, MSTVision introduced HALCON to the customer team for various reasons:

  • HALCON offers the highest scope of different technologies. This is particularly helpful if the required machine vision technologies are not yet defined.
  • HALCON provides a huge set of operators to solve all the different tasks required in the project.
  • HALCON can be used with a wide variety of hardware platforms: from compact vision sensors to high-performance PCs with multiple GPUs and FPGA frame grabbers.
  • HALCON’s powerful region concept is a huge advantage, when it come to the analysis of complex image content.
  • The combination of HDevelop with the HDevEngine enables a fast and flexible development process.
  • The software versioning and revision control system enables easy cross-team and cross-company development.

During the project, the debugging capability with the remote debugger was extensively used. Offline debugging for developers was essential. Each script stored all relevant input parameters. The dataset could then be transferred to the developer, who could then “import” these data into his HDevelop script. As a result, any problem on the machine could be identified and solved without the need of having access to the production equipment.

A huge part of the machine vision development process was conducted in HDevelop to make the best possible use of the high-grade IDE (Integrated Development Environment) for machine vision.

What the customer says

“MSTVision introduced us into the HALCON world and transferred tons of know-how to our team. Starting from how to use HDevelop, implementing scripts into C# using the HDevEngine and how to analyze issues in real time using the remote debugger. After this first project we are sure that HALCON will enable our team to solve future image processing tasks quickly and efficiently.”

Dr. Dietrich Drews, Manager Micro Assembly & Optics, HUBER+SUHNER Cube Optics AG

Text and images kindly provided by MSTVision GmbH.