High-Precision Alignment Servo of Planar Microlens Array.

Optical fiber is commonly used for long-distance communication. For short-distances, parallel communication should be developed, but this requires a multiport connection. A planar microlens array (PML) has been developed and is considered adequate for such an application. However, alignment of the PML is very difficult and is an obstacle to wide application. This paper introduces a new technique for aligning two PMLs using many-to-one optics. Instead of aligning each lens, optical focuses of PML are projected to one image by using many-to-one optics which consist of optical neurons. This image is input to a microcomputer, and a vision servo is developed. Two vision algorithms are tested : one calculates the alignment error from the digitized image, while the other uses a layered neural network. The alignment capability is tested and discussed.