Application of robust, high-accuracy positioning for autonomous ground vehicles - Robotics Institute Carnegie Mellon University

Application of robust, high-accuracy positioning for autonomous ground vehicles

Conference Paper, Proceedings of Unmanned Systems North America (AUVSI '04), August, 2004

Abstract

In this paper we review the need for robust, high-accuracy positioning for path guidance in the presence of large structures (such as buildings and trees) and for the integration of sensor data. Such capability is necessary in many applications that require precision over vast areas of operation. We find that unless integrated tightly, combined INS/GPS systems diverge significantly when line of sight to GPS satellites is lost. This results in a long recovery period especially when cm level positioning is required. High-accuracy inertial sensing is also useful when registering sensor data taken from a moving vehicle. This is particularly the case when data from a scanning sensor must be integrated over time to form a composite picture that aids in detecting obstacles that would hinder an autonomous vehicle. We show how a tight integration between inertial sensing and GPS positioning resolves some issues in the loss and reacquisition of satellites and how the inertial positioning is used to register sensor data. We present results from two applications of autonomous ground vehicles that use integrated GPS/INS systems to follow prescribed paths [8] and to sense the environment around them [11, 12]. One operates in smooth terrains, and the other is intended for cross-country operation.

BibTeX

@conference{Roth-2004-9011,
author = {Stephan Roth and Sanjiv Singh},
title = {Application of robust, high-accuracy positioning for autonomous ground vehicles},
booktitle = {Proceedings of Unmanned Systems North America (AUVSI '04)},
year = {2004},
month = {August},
}