next up previous
Next: The Action Model Up: Markov Localization for Mobile Previous: Implementations of Markov Localization

Metric Markov Localization for Dynamic Environments


In this section we will describe our metric variant of Markov localization. This includes appropriate motion and sensor models. We also describe a filtering technique which is designed to overcome the assumption of a static world model generally made in Markov localization and allows to localize a mobile robot even in densely crowded environments. We then describe our fine-grained grid-based representation of the state space and present techniques to efficiently update even large state spaces.

Dieter Fox
Fri Nov 19 14:29:33 MET 1999