DocumentCode :
759272
Title :
Simultaneous localization and map building of skid-steered robots
Author :
Anousaki, Georgia C. ; Kyriakopoulos, Konstantinos J.
Author_Institution :
Dept. of Mech. Eng., Nat. Tech. Univ. of Athens
Volume :
14
Issue :
1
fYear :
2007
fDate :
3/1/2007 12:00:00 AM
Firstpage :
79
Lastpage :
89
Abstract :
In this article, an alternative scheme is proposed for the SLAM problem, where the state vector holds only the robot pose and the map is feature-based in the form of line segments. It is a lighter representation in comparison to occupancy grids and point maps, something necessary as large scale environments are addressed. The basic idea is to obtain one estimate of the robot´s pose from an innovative dead-reckoning scheme and one from a laser scan matching algorithm. The fusion is done through a covariance intersection filter, avoiding the assumption that holds in Kalman filter that the cross correlations are zero. In this way, the uncertainty of pose estimate while the cross correlations are unknown and non zero are minimized. In the following section, the general framework of the SLAM algorithm is described. Then a short description of the dead-reckoning scheme, feature extraction scheme and the map matching algorithm are discussed. The loop with the covariance intersection fusion of the robot pose estimates are concluded. The description of the global map update algorithm and the results from representative experiments in real world experimental conditions are presented
Keywords :
SLAM (robots); feature extraction; filtering theory; mobile robots; road vehicles; Kalman filter; SLAM algorithm; covariance intersection filter; dead-reckoning scheme; feature extraction scheme; global map update algorithm; laser scan matching algorithm; line segments; map matching algorithm; robot pose estimates; simultaneous localization and map building; skid-steered robots; Filters; Land vehicles; Large-scale systems; Laser fusion; Mobile robots; Orbital robotics; Remotely operated vehicles; Robotics and automation; Simultaneous localization and mapping; Technological innovation;
fLanguage :
English
Journal_Title :
Robotics & Automation Magazine, IEEE
Publisher :
ieee
ISSN :
1070-9932
Type :
jour
DOI :
10.1109/MRA.2007.339625
Filename :
4141036
Link To Document :
بازگشت