Learning Landmarks for Robot Localization

Abstract

Our work addresses the problem of learning a set of visual landmarks for mobile robot localization. The learning frame-work is designed to be applicable to a wide range of envi-ronments, and allows for different approaches to computing a pose estimate. Initially, each landmark is detected using a model of visual attention and is matched to observations from other poses using principal components analysis. At-tributes of the observed landmarks can be parameterized us-ing a generic parameterization method and then evaluated in terms of their utility for pose estimation. We discuss the sta-tus of the work to date, and future directions. Problem Statement Our goal is to develop a framework for a robotic system which can automatically acquire knowledge of its environ-