DP-SLAM Release 0.1
Austin Eliazar
http://www.cs.duke.edu/~eliazar/
Ronald Parr
http://www.cs.duke.edu/~parr/
We are pleased to announce the availability of source code for DP-SLAM.
DP-SLAM is an algorithm for efficient Simultaneous Localization and
Mapping (SLAM). DP-SLAM produces extremely high resolution maps and
is accurate enough to map many domains with only a single pass over
the sensor data, and no explicit loop closing or correction routines.
Although this is a 0.1 release, it contains the most recent, as yet
unpublished, advances in the DP-SLAM framework.
The version of DP-SLAM used in this code has asymptotic complexity of
O(AP) per iteration, where A is the area swept out by your laser range
finder, and P is the number of particles assigned to DP-SLAM's
particle filter. This is the same asymptotic complexity as pure
localization with a known map. However, DP-SLAM does have a much
larger constant factor than pure localization and may require more
particles than pure localization. The most recent published version
of DP-SLAM has asymptotic complexity of O(AP^2).
The code also includes a new hierarchical mapping capability that
significantly reduces drift over long robot trajectories.
For details and code, please visit the DP-SLAM web page:
http://www.cs.duke.edu/~parr/dpslam/
--
Ron Parr, Assistant Professor
Department of Computer Science, Duke University, Durham, NC 27708-0129
Email: [EMAIL PROTECTED]
_______________________________________________
uai mailing list
uai@ENGR.ORST.EDU
https://secure.engr.oregonstate.edu/mailman/listinfo/uai
