Vision system puts head-up displays to the test

The HUD quality control system installed on the production lines of the BMW 5- and 6-Series is an example of top performance industrial image processing.

BMW clients expect the best and demand the most accurate equipment when buying a car from the Munich based manufacturer.

Increased demand for a "head-up display" (HUD) meant that BMW needed to find a quality control system for its HUDs.

This system would need to cope with the speed of their production line while maintaining high safety standards.

Now the HUD quality control system installed on the production lines of the 5- and 6-Series is an example of top performance industrial image processing.

The advantages of a HUD as an option in a car have created a rapid increase in demand.

The HUD is a driver information device, developed and produced by Siemens VDO Automotive.

The image-producing unit of the head-up display projects a virtual image onto the windscreen by means of an array of optical mirror systems.

The displayed image is projected in the viewing field of the driver at a distance of about 2m.

The driver can therefore perceive all important information regarding the vehicle and navigation within his/her normal field of viewing.

In this way, the driver has optimal access to information with minimum disruption of concentration.

If the HUD is even slightly misaligned, this will interfere with the driver's perception of the surrounding environment such street markings, vertical lines of houses, or garage entry.

That is why the imaging quality of the projected display in the vehicle is the decisive criterion for acceptance of the HUD by the user.

Automobile manufacturer BMW in Munich was concerned with the introduction of a head-up display as an optional equipment feature for the new car types of the 5- and 6-Series.

For this purpose, the department for Technical Integration at the Research and Innovation Centre of BMW in Munich needed a precise quality control system suitable for industrial use.

The fact that up to this point there was no appropriate quality control system based on image processing available meant the objective was to develop an absolutely new product within a very tight time frame.

Measuring methods for the control of imaging quality of HUDs had previously been developed in the laboratory, but there was still a long way to go to reach a solution for the production process.

These studies provided important target specifications for the preparation of the system design for this quality assurance project.

To add to the technical constraints, this quality control system would have to cope with the interaction of the windscreen and the projection of the HUD installed in the vehicle.

Even minor defects in a completely assembled vehicle would not be tolerated at the end of the assembly line.

On the basis of the system design specifications, BMW's Technical Integration (TI) Division contacted several suppliers of machine vision systems seeking a solution based on image processing which would work with the required quality assurance system of the head-up display.

In mid 2002, the company decided to work with Gefasoft, a partner system integrator (PSI) of Cognex.

The system solution demonstrated an advanced approach which convinced those responsible at BMW.

Measurement parameters had to be determined and then checked quickly and efficiently without interrupting production.

The timing frequency of the production line determined the time available for testing, and the complete testing system for the head-up display had to, of course, prove financially viable by functioning within a predefined time-frame and coping with the corresponding numbers of vehicles.

The testing system had to be suitable for both left- and right-hand-drive variants.

An additional requirement meant that the testing system of the HUD also would need to be fully traceable and embedded into the information infrastructure of the entire corporate network.

Ultimately, the testing system of the HUD has to provide an objective result regarding quality, evaluating and assessing with much greater precision and reliability than the user.

It was already clear in the feasibility study and during project planning, that this could only be handled with a high-performance PC-based image processing system.

Efficient hardware was required, and the algorithms of the vision software would be required to function with utmost reliability and precision.

No jams or standstills are tolerated on BMW's two-shift assembly lines.

The complete image acquisition and evaluation process of 50 acquired test images must be completed within a time-frame of less than two minutes.

The complete testing-equipment has to be swung into the vehicle as production is in progress, positioned exactly, secured in place, and then needs to be removed from the vehicle again.

This means a time-slot of approximately 30 seconds for image acquisition and image processing calculations in the test cycle.

The wide range of visual qualities of the projected virtual images make the evaluation of the images considerably more difficult.

Image processing cannot however be a cause for error in quality control under changing environmental production conditions.

This includes, among other issues, the problem-subject "contrast of displays" under varying conditions of illumination.

Vision-tools from the CVL (Cognex Vision Library) proved to be of great advantage.

They provide interactive access to different programming levels, in-depth vision-software, including an option for a more precise detail optimisation.

That is enabled the final solution - the selection of the tools, parameterisation, linking and detail optimisation - to be completed with a higher degree of safety and within a shorter development time.

Cognex vision tools perform to the highest of standards without compromising accuracy or safety.

The most important individual features in the object image, such as edges, measurements, shapes, angles, bends and shades, are specifically and separately identified.

The spatial conditions and relations between these central features of the acquired image are compared with the real-time image.

The position of the object is then precisely determined according to this information.

Certain features can therefore be recognised at a considerably higher level of reliability, precision and speed.

The influences of changing conditions of illumination and contrast are eliminated by a simultaneous investigation of the contour and the structure of the object image.

These qualities proved to be of advantage in the detection of deformations and distortions of the HUDs.

The functional safety of the vision tools also satisfied standards specified by BMW, which required it to be above 99%.

The image processing system must fulfil certain standards at BMW, and Cognex has for many years been among the suppliers who meet these requirements.

The only real issue of interest is the reliable operation of the complete system.

A stand-alone system could not be permitted on a globally parameterised production system.

The most important issue for those working at the plant was that the image processing be fully and reliably integrated into the plant diagnosis system.

The co-operation with Gefasoft proved to be of great benefit; the enterprise contributed greatly to the solution thanks to their flexibility and their capacity to react quickly within a limited schedule.

By 2004, the first head-up display testing systems were installed.

The positive results of the operation resulted in the head-up display testing systems also being installed in the normal series production in the factory at Dingolfing.

Today, approximately a dozen HUD testing systems are in operation for BMW around-the-clock, ensuring the faultless quality of the HUDs.

The test results are stored in a database in the central server, thus providing for total traceability.

The test-images are archived, and in case of customer requests, the flawless condition of the product at the time of delivery can be proven, even years later.

Back to top