Eigenspaces of Panoramic Images for Mobile Robot Localization

Matjaž Jogan (2002) Eigenspaces of Panoramic Images for Mobile Robot Localization. MSc thesis.

Preview

PDF Download (9Mb)

Abstract

Appearance--based visual learning and recognition techniques that are based on models derived from a learning set of 2D images are being widely used in computer vision applications. In robotics, they have received most attention in visual servoing and navigation. In this thesis we discuss a framework for visual self--localization of mobile robots using a parametric model built from panoramic snapshots of the environment. Particularly, we propose solutions to the problems related to robustness against occlusions, illumination and invariance to the rotation of the sensor. Our principal contribution is the ``eigenspace of spinning--images'', i.e., a model of environment which successfully exploits some of the specific properties of panoramic images in order to efficiently calculate the optimal subspace in terms of Principal Components Analysis (PCA) of a set of training snapshots without actually decomposing the covariance matrix. By integrating a robust recover--and--select algorithm for the computation of image parameters we achieve reliability even in the case when the input images are partly occluded or noisy. Further, by applying a bank of gradient filters, we achieve a significant level of insensitivity to changes in the illumination of the environment. In this way, the robot is capable of localizing itself in realistic environments.

Actions (login required)