Computer Science Capstone Design
Capstone Project Description, Spring 2008

Automated Geometric Centroiding System for the Alignment of the Navy Prototype Optical Interferometer

Sponsor Information:

Jim Clark, Mechanical Engineer,
Naval Research Lab, Navy Prototype Optical Interferometer
10321 W Observatory Rd,
Flagstaff, AZ 86001
928 773-4868, ext 203; jhc@sextans.lowell.edu

Joshua Walton, Mechanical Engineer,
Naval Research Lab, Navy Prototype Optical Interferometer
10321 W Observatory Rd,
Flagstaff, AZ 86001
928 773-4868, ext 209; jpwalton@sextans.lowell.edu

Project Overview:

The Navy Prototype Optical Interferometer (NPOI), located just outside Flagstaff, Arizona, consists of 170 precision flat mirrors mounted inside 9000 cubic feet of vacuum, positioned along a 400 meter diameter Y-shaped array. Quantifiable precision alignment of each mirror is critical to the efficient operation of the interferometer telescope system; light must be very accurately directed through the system. When the array was built, each mirror was precisely positioned. Over time, however, with natural thermal variations, the pointing of the mirrors drifts out of tolerance and requires correcting. Depending on temperature variations and other factors, this realignment process has to be undertaken as often as every night. This process is currently done manually, with the human eyeball and human operation of stepper motors to tweak the mirrors. Not only does this take much time and effort, it is also not as accurate as one could hope for; human visual systems are limited and "bump and check again" human adjustment of stepper motors is error-prone.

This project explores improvement of this process by replacing the human element with an automated system. In particular, we would like to explore using a CCD camera to perceive alignment, calculate corrections, and then send control signals to the stepper motors. The goal is to make the re-alignment process very fast, efficient and automated so that, even during nightly astronomical observations, it is possible to quickly check and re-adjust the mirrors in the optical train.

More generally, we would like to support a more diverse se of science programs and expansion of capabilities at the NPOI, which will require nightly reconfigurations of optical mounts within the vacuum system. As part of this process, efficient alignment of the optical train is required.

How the alignment system works: The basic concept is that alignment is by sighting a special LED target. A "pop-up" light emitting diode (LED) (mounted on a motorized target arm) is attached to each mirror mount and, when popped up, aligns perfectly to indicate the mirror’s center point. A focusable precision alignment telescope, mounted in a v-block assembly at the other (distant) end of the array galley, is employed as the basic alignment tool. Presently, the human eye acts as the primary detector and a subjective decision is made regarding the spatial error between the center of the LED target and the crosshairs of the telescope. The human eye detector is adequate for building and verifying rough alignments. However, the system is inefficient, slow to use, and requires much skill and effort to achieve optimal alignments, and is therefore inadequate for nightly reconfigurations and precision alignments.

We require the development of an automated imaging/centroiding system that has the following capabilities:

  • Quantify the error in alignments in near real time
  • Fast, efficiency and reliable
  • User-friendly interface
  • Adaptable to future expansions
  • Graphical user interface (GUI) such that a technician can operate the system and interpret the results
  • Expandable such that ultimately it will interface with computerized control electronics, i.e., will control the actal stepper motors, to form a fully automated alignment system.    

For more detailed information about the system and related alignment challenges, we include two recent papers submitted to recent SPIE conferences: Paper 1 and Paper 2. Both are Word documents.

Knowledge, skills, and expertise required for this project:
  • Project management skills
  • Computer programming, graphical user interface design
  • Image capture and noise suppression,
  • Geometric centroiding techniques and algorithms, calibration of plate scale, CCD and CMOS cameras, communications protocols.

Equipment Requirements:

The sponsor will provide the alignment telescope and mount, targets and target controls, digital camera(s) for mounting on the alignment telescope, frame grabber, and possibly a laptop for data logging.

Deliverables:
  • Software to compute and display the geometric alignment error. Specifically, we see three levels of increasing complexity:
    • Level 0: Bare minimum function. The system will capture a CCD camera image of the distant LED, then compute the alignment error. Basic proof of concept with rudimentary interface.
    • Level 1: Something actually usable. Elegant interface for daily use by end users. Include capability to generate analog signals to stepper motors, i.e., you take the reading, examine the error, hit the "apply corrections" button, and re-check.
    • Level 2: Extraordinary success: System designed as a web-accessible tool (i.e. could be done remotely) that could support an unlimited number of alignment systems, i.e., you select the light gallery you want to align, it has controls to pop up the LED and capture the image, then apply the corrections and double check the alignment.
  • A manual including instructions on how to use the software
  • A GUI such that a technician can intuitively use the system without also having intimate knowledge of it.
  • System checks (from the GUI) against standards (to be developed) in order to calibrate the system for target distance/focus variations.