Contactless Position Measurement System

[31-MAR-23] Our Contactless Positioner Measurement System (CPMS) is designed to permit automatic alignment of highly-reflecting, precision components that cannot or should not be touched by human hands until they are ready to be bolted together. In the case of Superconducting Radio Frequency (SRF) string assembly, human contact increases the chance of contamination, which ruins the performance of an SRF cavity. In the case of the components of a fusion reactor, these are likely to be radioactive, and so must be handled for as short a time as possible by humans. Measuring the location of highly-reflecting objects is challenging. The CPMS uses an infrared background to capture stereoscopic silhouette images of highly-reflecting metallic objects, which we then analyze to determine the location and orientation of these objects. The reflections of ambient light that make it so difficult to locate these objects are subdued by the use of an infrared background in a room in which the only ambient light comes from fluorescent or solid-state lamps. Development of the CPMS is funded by the United States Department of Energy (DoE) under Grant 13557088.

Figure: Two Contactless Position Measurement Systems on an SRF String Assembly. Each CPMS consists of two calibrated, infra-red only Silhouette Cameras (SCAMs), a calibrated reference plate, and a backlight to form the silhouette images.

The CPMS measures the position of two components in the coordinate system of its calibrated reference plate. Because the components are machined out of stainless steel, we will know their dimensions to within a ±125 μm (5 mils). By using its two images, and its knowledge of the component shapes and sizes, the CPMS provides a measurement of the relative position of the two mating surface that is accurate to better than 200 μm rms. These measurements allow us to maneuver the two components into contact using motorized stages, after which the two components are no longer vulnerable to contamination, and may be bolted together by hand. The silhouette cameras (SCAMs) are similar to BCAMs, but with larger field of view, and correspondingly lower precision. The SCAM mounts on three balls in exactly the same way as a BCAM. The reference plates operate in the same way as the alignment bars in the ATLAS End-Cap Alignment System, see here, only they are calibrated plates rather than calibrated bars.

Figure: Silhouette Image of a Steel Sphere on a Post. The image is taken in infrared. We see reflections of our windows in the sphere. Sunlight contains infrared. We see bright spots in the backlight, which are the LEDs shining straight through.

To measure the location of a component we begin with a silhouette image, guess the approximate location of the component, project a simulation of the component onto our image, and we compare the disagreement between the simlated image and the actual image. We adjust the position of the simulated component until we minimise the disagreement between the two images. The position of our simulated component is our measurement of the position of the actual component.

Figure: Classification Color Coding. Left: pre-fit. Right: after manual fit. Transparent overlay indicates agreement between projection and image.

We obtain ±100 μm precision with the above sphere and post in directions perpendicular to the SCAM axis, and ±300 μm parallel to the axis. When we convert to root means square precision, and scale the geometry by a factor of two to 100 cm range, our precision is consistent with our target of 200 μm rms in all direction.

Figure: Test Stand One, a Half-Size Stereo SCAM Breadboard. Object ranges 40-80 cm. A full-size would have objects at 80-160 cm.

In our Test Stand One (TS1), we equip our SCAMs with 25-mm focal length lenses and 3.7 mm apertures. Our backlight is 20 cm square. We focus the cameras upon our object at range 75 cm. With 50-ms exposure, we we obtain ±200 μm precision from ranges 58 cm to 88 cm. The precision at the extreme ranges is an artifact of our image analysis procedure. We classify pixels as either silhouette or non-silhouette. When the image becomes blurred, our classification is affected by our exact choice of intensity threshold. We are working on improving the classification's performance in the presence of defocus. Our projection routine is inefficient: one projection of sphere and cylinder takes one second on our 1-GHz laptop. Automated fitting of the object position takes roughly one minute. Over the next few months, we will be working on improving the efficiency of the projection. The depth of field of the SCAMs will be improve as we decrease the aperture. If we halve the aperture diameter, we double the depth of field. At the same time, however, we quadruple the exposure time. We are designing a backlight that will provide steady illumination for exposures up to one second.

Development Log: Development of the CPMS at OSI starting August 2022.

Contactless Position Measurement for Highly Reflective Component: Project abstract and summary from our Small Business Innovation Research (SBIR) application to the Department of Energy (DoE). Submitted 26-FEB-22, awarded 27-JUN-22, grant number 13557088.

Press Release 31-AUG-22: Company's public announcement of receipt of award.

Infrared Backlight (A3046): A diffuse source of infrared light that may be flashed for bright, sharp, silhouette images.

Boston CCD Angle Monitor (BCAM): Home page of the metrology cameras we use to construct the CPMS global coordinate system.