Active Markov Localization for Mobile Robots
Dieter Fox

To navigate reliably in indoor environments, a mobile robot must know
where it is.  Over the last few years, there has been a tremendous
scientific interest in algorithms for estimating a robot's location
from sensor data.  Most existing localization approaches are passive,
i.e., they do not exploit the opportunity to control the robot's
effectors during localization.  In this talk we will introduce an
active localization approach based on a geometric variant of Markov
localization.  The approach provides rational criteria for (1) setting
the robot's motion direction (exploration), and (2) determining the
pointing direction of the sensors so as to most efficiently localize
the robot.  Furthermore, we will propose another extension of Markov
localization which significantly improves the robustness of position
estimation even in densely populated environments.

The experiments given in this talk demonstrate that our implementation
of Markov localization in combination with these extensions is able to
(1) robustly estimate the position of a mobile robot even in densely
populated environments, (2) detect localization failures, and (3)
autonomously (re-)localize a mobile robot from scratch.