Automatic illumination planning for optical quality control could further enhance quality and reliability

Cameras are being used in optical quality control to an in creasing ex-tent. In many cases, this approach proves more reliable than the manual testing of engines or individual components carried out by humans. Key factors are precisely calculated positions for the camera and light sources, for which the DaimlerChrysler researchers have developed a mathematical planning tool. Up to now, the correct arrangement of the lights and cameras, their positions and internal parameters - objective lenses, exposure times and aperture settings - called for a great deal of experience and sensitivity. Only once a production cell was fully set up could the last steps in optimization be taken. With their newly developed method, the DaimlerChrysler researchers are now able to take all parameters into account as early as the planning phase of a production cell, allowing them to save both on components and on valuable space.

Enlightening moments

As many cameras as are necessary should be used, but no more: the system must operate rapidly, precisely and economically. At some workplaces - such as final engine inspection - space is often so limited that there is room for only a few cameras, or just one. The DaimlerChrysler researchers adopted a holistic approach to solving the complex requirements placed on optical quality control. Oriented towards human work processes, they proceeded in four stages: depending on the inspection plan and the component's geometry, the basis of calculation was first of all determined. In the second phase, the approximate camera configuration and spatial distribution around the test object was estimated. The optimized positions for the cameras and lights were determined in the next phase. Finally, the appropriate settings for each camera such as focus, exposure and shutters were calculated. The result was a complete test station at which the number of cameras, the settings and lights stood perfectly matched.

Interesting as it may sound, the automatic planning process mentioned above is much more efficient than the human approach, as it dispenses with the time-consuming constant testing and matching of all settings and individual steps leading up to the optimal configuration. For the purposes of calculation, it is irrelevant whether the cameras and lights are fixed or are newly positioned by robots. The test station would thus be highly versatile and can accommodate the most diverse test procedures.

Reflective metal

In addition to the geometry of a component, the researchers had to take into account the material used. In the case of metal components, the challenge lay in the strongly reflective surfaces. Each metal reflects light in its own distinctive manner. For the camera, this reflection characteristic plays an important role because its settings are directly dependent on the amount of light energy received per pixel. This even indirectly affects the depth of focus: a low level of exposure yields a small depth of focus. In such cases, as a result of the material characteristics the calculations could require an additional camera to examine a certain area of the component.

Final engine inspection

Whether it be the geometry or the material - the DaimlerChrysler researchers had no easy job when it came to quality inspection of internal combustion engines: the engine consists firstly of rigid components such as seal caps, plugs, sockets, bolts and flanges that the system must identify; and then there are the moving components such as hoses and cables that until now had to be tested manually. A perfectly functioning but loose cable could be damaged when the engine is running. The automatic testing system therefore had to identify the three-dimensional position of the cable in relation to the entire engine. As soon as the computer compared the data with the design plans, it could reliably determine whether the engine passed the test. From engine, metal and paint testing via welding seams and sheet metal, up to distinguishing pores and shrinkage cavities from simple impurities - the DaimlerChrysler researchers enabled computers to identify and reliably analyze three-dimensional objects. The new optical test method would thus contribute to even higher production quality in future.