Progress in 3-D Mapping and Localization - Robotics Institute Carnegie Mellon University

Progress in 3-D Mapping and Localization

Conference Paper, Proceedings of 9th International Symposium on Intelligent Robotic Systems (SIRS '01), pp. 145 - 154, July, 2001

Abstract

This paper is a summary of results obtained in the past few years in the area of 3-D mapping and robot localization. The emphasis of this work is the reconstruction of three-dimensional representations of the environment from sensor information, assuming inaccurate or absent robot pose information, and assuming general 3-D configurations, e.g., not limited to 2-1/2D elevation maps. The approaches are divided into two broad classes: Surface matching, in which large pieces of 3-D surfaces are matched across observations in order to recover the transformations between observations, and feature matching in which individual features extracted from the input data are matched. Surface matching is most applicable to robots equipped with range sensors such as stereo or ladar, while feature matching is most applicable to video-based systems. We report on experiments and applications in the areas of terrain map building, modeling of interior environments, modeling of individual objects from many views, cooperative stereovision using teams of robots, and simultaneous recovery of structure and motion from bearing-only sensors.

BibTeX

@conference{Hebert-2001-8286,
author = {Martial Hebert and Matthew Deans and Daniel Huber and Bart Nabbe and Nicolas Vandapel},
title = {Progress in 3-D Mapping and Localization},
booktitle = {Proceedings of 9th International Symposium on Intelligent Robotic Systems (SIRS '01)},
year = {2001},
month = {July},
pages = {145 - 154},
}