Robot Pose Estimation Using Laser Range Finder with Iterative Closest Point Algorithm in an Unknown Stationary Environment
Date Issued
2010
Date
2010
Author(s)
Yin, Ming-Tzuoo
Abstract
Localization is an important issue for robot navigation in an unknown stationary environment. ICP (Iterative Closest Point) algorithm can be used to solve problem of curve registration by the concept of least square errors. It is a useful tool for mapping matching, localization, and map building because of its capability to eliminate systematic and nonsystematic errors of robot simultaneously.
However, differences in two mappings and restrictions from laser range finder make complete mapping matching impossible in actual situation. In this thesis, a method is proposed to find common parts of the original mappings called hybrid potential common parts before using ICP algorithm. Then, there are three methods to promote accuracy and robustness of overall algorithm. The experimental difficulties in reality are mentioned in addition.
The goal of above discussed methods is to assist ICP algorithm to match maps more successfully and overcome different types of obstacles in an indoor stationary environment.
Subjects
Localization
ICP algorithm
nonsystematic errors
laser range finder
pose estimation
Type
thesis
File(s)![Thumbnail Image]()
Loading...
Name
ntu-99-R97921052-1.pdf
Size
23.32 KB
Format
Adobe PDF
Checksum
(MD5):657c023bf55183abff984250eca847d3