AASS Seminar - Extended Representation of Uncertainty in Active SLAM

The research centre AASS arranges a seminar with Joachim Clemens, University of Bremen.

Abstract

Localization, mapping, and planning are three of the most fundamental tasks in mobile robotics. In all three areas, it is essential to correctly handle and represent the inherent uncertainty of the processed information. Sensor data used to localize the robot and build a map of the environment are usually noisy as well as incomplete and may conflict each other. This needs to be considered during the fusion process and has to be reflected by the result. The latter is in turn used to decide on the further course of action, where the uncertainty has to be taken into account as well. We consider an extended representation of uncertainty in the context of active SLAM. In particular, an evidential approach to SLAM, path planning, and active exploration is presented, where belief functions are used to model the uncertainty in the estimated map. In addition to the algorithm, we present empirical results and compare the approach to a probabilistic one.

Speaker's bio

Joachim Clemens received the Engineer degree in computer engineering from the Hochschule Niederrhein University of Applied Sciences, Krefeld, Germany, in 2008 and the Master degree in computer science from the University of Bremen, Bremen, Germany, in 2011.

He is currently working towards his Ph.D. within the Cognitive Neuroinformatics group, University of Bremen, Bremen, Germany, headed by Kerstin Schill. His research interests include simultaneous localization and mapping for robotics, space applications, and highly automated driving as well as extended uncertainty representations.