Biologically Inspired Monocular Vision Based Navigation and Mapping in GPS-Denied Environments Koray Celik * , Soon-Jo Chung † , Matthew Clausman ‡ , and Arun K. Somani § * Iowa State University, Ames, Iowa, 50011, USA This paper presents an in-depth theoretical study of bio-vision inspired feature extrac- tion and depth perception method integrated with vision-based simultaneous localization and mapping (SLAM). We incorporate the key functions of developed visual cortex in several advanced species, including humans, for depth perception and pattern recognition. Our navigation strategy assumes GPS-denied manmade environment consisting of orthog- onal walls, corridors and doors. By exploiting the architectural features of the indoors, we introduce a method for gathering useful landmarks from a monocular camera for SLAM use, with absolute range information without using active ranging sensors. Experimental results show that the system is only limited by the capabilities of the camera and the availability of good corners. The proposed methods are experimentally validated by our self-contained MAV inside a conventional building. I. Introduction Undoubtedly the most influential perceptual sensory mechanism in biology is vision. Contrary to popular belief, even echolocating bats rely on vision for ranging beyond the attenuation of their sonar, with the visual acuity to discriminate a coin from several meters. 1 Perhaps the most interesting aspect of vision is the ability to estimate the range to an object without emitting any wave signal that can be detected by the prey; the ultimate eavesdropping tool. These properties render vision a particularly useful method for situation awareness in predators, and, an intuitive for SLAM in probabilistic robotics. A vision guided platform also has a wide potential for military applications held at GPS denied environments. Since the photoreceptor cells in retina capture the surrounding geometry through photometric effects, the work of merit in visual ranging belongs to the hyper-complex neurons in visual cortex. The intricate details as to how these neurons function is still amystery. However, studies such as Hubel et al. 2 shed light on how visual cortex might operate, in which the extracellularly responses from the dorsal aspect of cat brain was studied using the actions of the animal as a probe to estimate the functions of these neurons in response to visual stimuli. It was discovered that the visual cortex prefers tracking small contours of the environment, and the optokinetic nystagmus of the animal suggested that moving contours were of particular interest. Our approach takes these contours into account, inspired by animals with two dimensional retinae that perceive depth via such monocular visual cues such as line perspectives, relative height, texture gradient, and motion parallax. We would like to stress the term monocular here; studies on cats have shown that 80% of all cells in visual cortex were influenced independently by the two eyes, suggesting that when it comes to long-range navigation, monocular vision is more influential than stereo-vision in biology. Eagles for instance, utilize the two eyes independently to track multiple landmarks (and targets) simultaneously, and estimate depths in a monocular manner. Using both eyes in unison is only useful for objects within immediate vicinity. All other times they stare in parallel to obtain a single wide-angle image. Similarly in probabilistic robotics, stereo vision does not have the potential for real-time online SLAM applications with * Doctoral Research Assistant, Department of Electrical and Computer Engineering, koray@iastate.edu. † Assistant Professor of Aerospace Engineering and Electrical & Computer Engineering, sjchung@alum.mit.edu. ‡ Department of Electrical and Computer Engineering, mclausma@iastate.edu § Anson Marston Distinguished Professor, Jerry R. Junkins Endowed Chair, and Department Chair, Department of Electrical and Computer Engineering, arun@iastate.edu. 1 of 17 American Institute of Aeronautics and Astronautics AIAA Infotech@Aerospace Conference <br>and <br>AIAA Unmanned...Unlimited Conference 6 - 9 April 2009, Seattle, Washington AIAA 2009-1962 Copyright © 2009 by the American Institute of Aeronautics and Astronautics, Inc. All rights reserved.