Radar obstacle detection: finding moving targets in noisy range data - Robotics Institute Carnegie Mellon University

Radar obstacle detection: finding moving targets in noisy range data

Conference Paper, Proceedings of 19th IEEE Instrumentation and Measurement Technology Conference (IMTC '02), Vol. 1, pp. 145 - 150, May, 2002

Abstract

Short-range (/spl sim/2-7 meter) integrated radar units in the 24 GHz (wavelength 1.2 cm) regime are now available as production prototypes, and are soon expected to be very inexpensively available to large commercial users (e.g., automobile manufacturers). We recently had an opportunity to examine the applicability of several pilot units as potential components in a "situational awareness" sensor suite for automobiles in urban traffic. The system goal is to detect, distinguish, and appropriately respond to nearby vehicles (cars, bicycles, etc.), life forms (pedestrians, dogs, etc.), and movable and fired features of the environment (trash cans, lamp posts, etc.). The question is whether the first generation radar prototypes, with all the warts typical of pioneering technologies, are nevertheless useful complements to an existing system that employs relatively mature sensor technologies, e.g., panoramic video analysis, stereo image analysis, laser rangefinding, ultrasonic range finding, integrated by an existing and reasonably well developed sensor fusion architecture. Our answer, based on a few experiments with three prototype units arranged in two geometries, is a definite maybe.

BibTeX

@conference{Siegel-2002-122422,
author = {M. Siegel and R. MacLachlan},
title = {Radar obstacle detection: finding moving targets in noisy range data},
booktitle = {Proceedings of 19th IEEE Instrumentation and Measurement Technology Conference (IMTC '02)},
year = {2002},
month = {May},
volume = {1},
pages = {145 - 150},
}