@InProceedings{pfaff08icra,
  title     = {Gaussian Mixture Models for Probabilistic Localization},
  author    = {Patrick Pfaff and Christian Plagemann and Wolfram Burgard},
  booktitle = {Proc.~of the IEEE Int.~Conf.~on Robotics \& Automation (ICRA)},
  address   = {Pasadena, CA, USA},
  year      = {2008},
  abstract  = {Range sensors have become popular for mobile robot localization since they directly measure the geometry of the local environment. In situations in which the robot operates close to obstacles or in highly cluttered environments, however, small changes in the pose of the robot can lead to completely different geometries measured by the range sensor. The resulting enormous variances in the likelihood of observations can lead to major problems in probabilistic approaches such as Monte Carlo localization as important hypotheses or particles might get lost which substantially decreases the robustness of such approaches. A common solution is to artificially smooth the likelihood function or to only integrate a small fraction of the measurements. In this paper we present a more fundamental and robust approach which models the likelihood function for single range measurements as a mixture of Gaussians. In practical experiments we compare our approach to previous methods and demonstrate that it provides a substantially more robust localization.},
  note      = {to appear},
  pdfurl    = {http://www.informatik.uni-freiburg.de/~plagem/pfaff08icra.pdf}
}