Probabilistic Robotics Navigation

Sebastian Thrun

PROBABILISTIC ROBOTICS
Nursebot Pearl, serving as an intelligent reminder and navigation aid for older adults, incorporates pervasively probabilistic navigation and user interaction software (developed jointly by Carnegie Mellon University, the University of Pittsburgh, and the University of Michigan).

Planning and navigation algorithms exploit statistics gleaned from uncertain, imperfect real-world environments to guide robots toward their goals and around obstacles.

I
52
CARNEGIE MELLON UNIVERSITY March 2002/Vol. 45, No. 3 COMMUNICATIONS OF THE ACM

n the past decade, the field of robotics has made notable progress in terms of flexibility, autonomy, and human interaction. Robots had been largely confined to factory floors and assembly lines, bound to perform the same narrow tasks over and over again. A recent series of successful robot systems has, however, demonstrated that robotics has advanced to where it is ready to conquer many new

fields, including space exploration, medical applications, personal services, entertainment, and military support. Many of them are highly dynamic and uncertain. Uncertainty arises for many reasons, including the inherent limitations of a particular model of the world, the noise and perceptual limitations in a robot’s sensor measurements, and the approximate nature of many algorithmic solutions. Finding mechanisms for handling this uncertainty is a primary challenge faced by robotics research today. Figure 1 shows three examples of successful robot systems operating in uncertain environments: a commercially deployed autonomous straddle carrier [3]; an interactive museum tour-guide robot [7, 11]; and a prototype robotic assistant called Nursebot for the elderly. The straddle carrier, developed at the University of Sydney, is capable of transporting containers faster than trained human operators. The tour-guide robot, one in a series, safely guides visitors through densely crowded museums, including...