The CMU Robotics Institute has released a new technical report by Stella X. Yu entitled, Computational Models of Perceptual Organization. The paper addresses the organization of sensory input into coherent and interpretable perceptual structures. Useful machine vision for robots requires low-level processes to reduce the computational complexity for higher level processes. But knowledge of the high-level processes is needed to guide the low-level processes away from noise. Yu thinks the solution to this chicken-and-egg problem lies in spectra graph theory. The paper is the result of Yu's work at the Active Perception Laboratory.