D. Fox, W. Burgard, F. Dellaert, and S. Thrun

MonteCarlo Localization: Efficient Position Estimation for Mobile Robots

Proc. of the Sixteenth National Conference on Artificial Intelligence (AAAI'99)


 

Abstract

This paper presents a new algorithm for mobile robot localization, called Monte Carlo Localization (MCL). MCL is a version of Markov localization, afamily of probabilistic approaches that have recently been applied with greatpractical success. However,previous approaches were either computationally cumbersome (such asgrid-based approaches that represent the state space by high-resolution 3D grids), or had to resort to extremelycoarse-grained resolutions. Our approach is computationally efficient while retaining the ability to represent(almost) arbitrary distributions.  MCL applies sampling-based methods forapproximating probability distributions, in a way that places computation``whereneeded.''  The number of samples is adapted on-line, thereby invokinglarge sample sets only when necessary.  Empirical results illustrate thatMCL yields improved accuracy while requiring an order of magnitude less computationwhen compared to previous approaches. It is also much easier to implement.

Download

Full paper [.ps.gz](1102 kb, 7 pages)

Bibtex

@INPROCEEDINGS{Fox99Mon,
  AUTHOR    = {Fox, D. and Burgard, W. and Dellaert, F. and Thrun, S.},
  TITLE     = {Monte Carlo Localization: Efficient Position Estimation for Mobile Robots},
  YEAR      = {1999},
  BOOKTITLE = {Proc.~of the National Conferenceon Artificial Intelligence}
}