IEEE ICRA 2012: Workshop on Stochastic Geometry in SLAM
Meeting Room 6, Saint Paul River Center,
175 West Kellogg Blvd, Saint Paul, Minnesota 55102, USA.
Friday 18. May 08.45 - 17.15.
Recent techniques from the
field of Stochastic Geometry have introduced the concept of a Random
Finite Set (RFS) representation of a multi target state, in
multi-target tracking. Representing the feature based SLAM map state as
an RFS, rather than the conventionally used random vector, is not
merely a triviality of representation. Recent research has shown that
Finite Set Statistics (FISST), developed for data fusion and estimation
with RFSs, when applied to SLAM, eliminates the necessity of fragile
map management and feature association algorithms. The RFS map concept
therefore provides a robust paradigm under which the true number of
features, which have entered the field(s) of view of an autonomous
vehicle’s sensor(s), as well as their locations, can be jointly
estimated in a Bayes optimal manner, while taking into account feature
detection and false alarm probabilities. Further, estimation has little
meaning without a concise notion of estimation error. Performance error
metrics for SLAM will therefore be introduced, which allow SLAM map
estimation error to be evaluated in its entirety.
The workshop will encompass presentations on the direct application of
FISST to SLAM with an introduction to FISST, SLAM solutions in the
presence of high levels of clutter in complex environments,
multi-vehicle SLAM, extended target tracking and adaptive information
retrieval based on visual sensors.
||Martin Adams & Ba-Ngu-Vo, Welcome and Workshop Introduction
||Ronald P. Mahler, "Finite-Set Statistics and SLAM" (Slides Presented at Workshop)
||Ba-Ngu Vo, "Stochastic Geometry and Bayesian SLAM"
||Daniel E. Clark, Chee S. Lee and Sharad Nagappa, "Single-Cluster PHD Filtering and Smoothing for SLAM Applications"
||Karl Granström, Christian Lundquist, Fredrik Gustafsson and Umut Orguner, "On Extended Target Tracking Using PHD Filters"
||John Mullane, Samuel Keller and Martin Adams "Random Set Versus Vector Based SLAM in the Presence of High Clutter"
|Diluka Moratuwage, Ba-Ngu Vo, Sardha Wijesoma and Danwei Wang, "Extending the Bayesian RFS SLAM Framework to Multi-Vehicle SLAM"
|Philip Dames, Dinesh Thakur, Mac Schwager and Vijay Kumar, "Adaptive Information Gathering Using Visual Sensors"
|Panel Discussion, "The Future of Stochastic Geometry in Robotic Navigation"