Adjustment of Xenon car head lamp with vision and Stewart platform. (Hexapod)

Xenon lamp The market share of the so called blue Xenon car headlamps is still increasing and chase away the 'old' Halogen headlamps .
The Xenon headlamps do have a better performance, the efficiency and the intensity are better then the Halogen lamps.
The Xenon headlamp does have a smaller light source and therefore a better focusing of light is feasible.

The light source itself is a plasma bow between two electrodes.
Because the customers does not want to adjust the lamp after replacement the position of this plasma bow with respect to the lamp holder is very tight tolerated.

Image top view Image side view For adjustment of a Halogen lamp the position of the filament can be measured and subsequent adjusted. In contrast to Halogen lamp a Xenon lamp doesn't have a filament but two electrodes. When a Xenon lamp is 'burning' the plasma bow in vertical direction does have a bow shape. (see images) The heat within the lamp and the gravity are responsible of the bow shape.

The centre of gravity of the plasma bow drifts as a result of the temperature so adjusting of this plasma bow is done when the lamp is 'burning'. A reliable and correct measurement of the plasma bow is possible after a burning time of at least 60 seconds. The temperature of the lamp is almost steady and so the plasma bow does not drift.

Vision and Stewart platform set up.

Adjustment set up Hexapod For measuring the plasma bow in 3D two perpendicular positioned cameras are used.
The telecentric lenses are equipped with Neutral Density filters to reduce the intensity of the Xenon lamp (see th schematically representation). For accurate displacement of the Xenon burner a Stewart platform is used also known as a Hexapod. On the platform a pneumatic gripper is mounted. This V-shaped gripper grasps the top of the burner.
After the lamp holder is accurate mechanical measured in 5D (no rotation measurement on lamp axis) the set point of the plasma bow is calculated.
The cameras measure the bow center of gravity and the Stewart platform is activated by the vision system to position the bow on previous calculated set point. For a minimum of control time the vision system controls the Stewart platform itself.

After the first measurement and manipulation of the Stewart platform a second measurement is done. Also the last correction is made by activating the Stewart platform. Next step is the fixation of the lamp by laser welding and after the gripper is released a last burner measurement is performed.

Vision and manipulator Algorithm

The vision system starts acquiring two images after a start measure signal is given. After acquiring two images of the plasma bow first the 'burning' check is performed. Image & drawings side view No plasma bow results in a rejection of the lamp. Next step is a rough detection of the bow itself followed by an accurate measurement of the bow shape and intensity. The shape of the bow is not always perfect and of course it changes constantly.
Important and difficult is the correct detection of the two hot spots at the electrodes. These hotspot define the position of the plasma bow in one direction. In some situations the detection of the hotspot is not reliable so 2 other bow-end detection techniques are used.
The center of gravity of the bow in the centre of the two hot spot define the position of the bow in the perpendicular direction. Because of some disturbances of the plasma bow in the image and the quarts glass in front of the plasma bow several corrections have been made.
Control diagram When the centre of gravity of the bow in two direction is measured for both cameras the exact world position of the bow with respect to the camera positions are calculated. Therefore a previous calibration of the cameras is performed.
Knowing the bow centre of gravity in world coordinates and the set point of the bow required position in also world coordinates, the correction of the gripper on the Stewart platform can be calculated.
Or after a correction calculation the Stewart platform can be manipulated. And of course the Stewart platform is previously calibrated.

The whole system, measurement of the lamp holder, the two cameras and the Stewart platform makes use of 3 different oriented system of coordinates. This makes the whole system complicated.
Xenon_lamp Xenon with electronics philips At the third and last measurement not only the final position of the plasma bow is measured but also 7 quality characteristics are measured. These 10 quality characteristics are send to an overall quality system and if one of the characteristics exceeds the quality limits, the lamp is rejected.
Errors reports including error codes and messages are given.
The cycle time of lamp adjustment is less then 3 seconds. The vision measurement accuracy is better then (standard deviation) 10 micron.