Academic Company Events NI Developer Zone Support Solutions Products & Services Contact NI MyNI

Customer Solutions

CNN Technology Uses NI Vision Products for Automotive Inspection

Author(s):

Fok Kah Weng, CNN Technology

Industry:

Automotive

Product:

LabVIEW, PXI/CompactPCI, Vision

The Challenge:

Developing an automated vision inspection system for vehicle instrument clusters.

The Solution:

Using National Instruments LabVIEW, PXI, and Vision to enhance the accuracy of the entire production test.


Building an Automated System

We received a request from a renowned Malaysian automotive contract manufacturer to develop an automated vision inspection system. The device under test (DUT) was a vehicle dashboard instrument cluster. Due to the complexity of the cluster, manual inspection was inefficient and time consuming. We determined that an automated inspection system would enhance the accuracy and productivity of the entire production test.

The DUT was a vehicle instrument cluster that included the following items:

  • Tachometer
  • Speedometer
  • Temperature gauge
  • Fuel tank gauge
  • In-vehicle indicators
  • Odometer
  • Clock

Creating a Test System with NI Hardware and Software

We used NI hardware and software in the entire test system. We selected the NI PXI system as the test platform, which included a PXI 8–slot chassis with an embedded Pentium MMX 233MHz PXI controller. We used the NI PXI-1409 high-quality image acquisition module, which connected to four units of NTSC industrial grade color CCD cameras.   We also used the controller serial port to communicate with a custom-built circuit buffer board for the EEPROM programming, which resided in the cluster storing calibration data for the cluster.   In addition, we used NI LabVIEW as the development software with the help of NI-IMAQ driver software and NI Vision image processing software.

We had the following inspection requirements:

  • Label inspection: The first inspection requirement, we were required to ensure the tachometer and speedometer label were correctly placed for the different DUT models. On the image processing function, the software obtained a reference point setting and used a pattern matching algorithm to match with a previously learned template.
  • Meter and gauge inspection: This was the major and crucial test requirement. The test software had to detect the meter’s pointer position and ensure it was within the required tolerance. We did not test all the positions on the meter. We only tested software at different calibration points, such as 90km/h and 120km/h points. In order to perform this, the software communicated with the respective gauge via serial port to command the gauge to move the pointer to the calibration point. The software then read the pointer position, referring to a reference point. Although the meter and gauge VIs provided by the NI Vision software helped the system perform the task, we implemented a series of image enhancement and morphology functions as well as ROI definition to ensure accuracy.
  • In-vehicle indicators inspection: There were more than 20 different indicators, such as high-beam and left/right signal indicators.
  • Odometer and clock inspection: The display for the clock and odometer was in a seven-segment digital display format. The software used a line profile and peak-valley detection algorithm for the image analysis.

Developing a Simple, Cost-Effective System

We developed the software with a simple, user-friendly operator graphical user interface. The display shows the UUT image, start and stop button, and standard production test statistical results. The software consists of a configuration module with the ROI setup utility, tolerance setup function, template image learning, and store retrieve capability by test model. It is flexible and easy to configure different test models.

Furthermore, we designed a calibration module to perform gauge position and color calibration to cater to material variances in different production lots. During the development phase, we took various measures into consideration to minimize the impact of light source brightness and object position variances, in order to ensure the robustness of the system. Using NI hardware and software, we developed a very cost-effective, automated vision inspection system.

For more information, please contact:

Fok Kah Weng

Software Manager

CNN Technology (M) SDN BHD

737-1-10, Kompleks Sri Sungai Nibong

Jalan Sultan Azlan

11900 Penang

Malaysia

Fax: 604-6448181

E-mail: fok@cnn.com.my