TY - GEN
T1 - Simultaneous Localization and Mapping (SLAM) based on pseudolinear measurement model with a bias reduction approach
AU - Pathiranage, Chandima Dedduwa
AU - Watanabe, Keigo
AU - Izumi, Kiyotaka
PY - 2007/12/1
Y1 - 2007/12/1
N2 - This paper describes an improved solution to mobile robot localization and map building problem based on pseudolinear measurement model through bias reduction approach. Accurate estimation of vehicle and landmark states is one of the key issues for successful mobile robot navigation if the configuration of the environment and initial robot location are unknown. Direct linearization of nonlinear models has always come up with information loss of original nonlinear models. A state estimator which uses linearized models can be a solution to a problem where high accuracy is not expected. A state estimator which can be designed to use the nonlinearity as it is coming from the original model has always been invaluable in which high accuracy is expected. Thus to accomplish the above highlighted point, pseudolinear measurement model is directly applied to the Kalman filter equations without losing the original nonlinearity. We address the pseudolinear bias problem through simple geometry translation of system states. This system is simulated using Matlab for vehicle-landmarks system and results show that the new approach performs much accurately compared to that of well known Extended Kalman filter (EKF).
AB - This paper describes an improved solution to mobile robot localization and map building problem based on pseudolinear measurement model through bias reduction approach. Accurate estimation of vehicle and landmark states is one of the key issues for successful mobile robot navigation if the configuration of the environment and initial robot location are unknown. Direct linearization of nonlinear models has always come up with information loss of original nonlinear models. A state estimator which uses linearized models can be a solution to a problem where high accuracy is not expected. A state estimator which can be designed to use the nonlinearity as it is coming from the original model has always been invaluable in which high accuracy is expected. Thus to accomplish the above highlighted point, pseudolinear measurement model is directly applied to the Kalman filter equations without losing the original nonlinearity. We address the pseudolinear bias problem through simple geometry translation of system states. This system is simulated using Matlab for vehicle-landmarks system and results show that the new approach performs much accurately compared to that of well known Extended Kalman filter (EKF).
UR - http://www.scopus.com/inward/record.url?scp=51549119057&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=51549119057&partnerID=8YFLogxK
U2 - 10.1109/ICIINFS.2007.4579151
DO - 10.1109/ICIINFS.2007.4579151
M3 - Conference contribution
AN - SCOPUS:51549119057
SN - 1424411521
SN - 9781424411528
T3 - ICIIS 2007 - 2nd International Conference on Industrial and Information Systems 2007, Conference Proceedings
SP - 73
EP - 78
BT - ICIIS 2007 - 2nd International Conference on Industrial and Information Systems 2007, Conference Proceedings
T2 - ICIIS 2007 - 2nd International Conference on Industrial and Information Systems 2007
Y2 - 9 August 2007 through 11 August 2007
ER -