A key thrust of NASA's Origins program is the development of
astronomical interferometers. Pursuing this goal in a cost-effective and expedient manner from the ground has led NASA to develop the Keck Interferometer, which saw first fringes between the twin 10m Keck telescopes in March of 2001. In order to enhance the imaging potential of this facility, and to add astrometric capabilities for the detection of giant planets about nearby stars, four 1.8 m 'outrigger' telescopes may be added to the interferometer. Robust performance of the multi-aperture instrument will require precise alignment of the large number of optical elements found in the six optical beamtrains spread about the observatory site. The requirement for timely and reliable alignments dictated the development of an automatic alignment system for the Keck Interferometer. The autoaligner consists of swing-arm actuators that insert light-emitting diodes on the optical axis at the location of each optical element, which are viewed by a simple fixed-focus CCD camera at the end of the beamtrain. Sub-pixel centroiding is performed upon the slightly out-of-focus target spots using images provided by a frame grabber, providing steering information to the two-axis actuated optical elements. Resulting mirror-to-mirror alignments are good to within 2 arcseconds, and trimming the alignment of a full beamtrain is designed to take place between observations, within a telescope repointing time. The interactions of the autoaligner with the interferometer delay lines and coude trains are discussed in detail. The overall design of the interferometer's autoaligner system is presented, examining the design philosophy, system sequencing, optical element actuation, and subsystem co-alignment, within the context of satisfying performance requirements and cost constraints.