Input combination for Monte Carlo Localization

Information Technologies - Applications and Theory 2009.

计算机科学;

David Obdr·z¶alek

Others  :  http://ceur-ws.org/Vol-584/paper7.pdf
PID  :  42229

学科分类：计算机科学（综合）

来源: CEUR

【 摘 要 】

One of the basic skills for an autonomous ro-bot is the ability to determine its own position. There are numerous high-level systems which provide precise position information, but the same task may be also solved using less advanced and less accurate sensors. In our paper, we show how a good output may be acquired from not so good inputs if it is combined using modi¯ed Monte Carlo Localization (MCL) system. The combination of more inputs also helps to acquire plausible results even in situations, where adding new sensor(s) to an existing system raises doubts about the position calculation, because the newly added sensor may provide di®erent position information (or data from which the position is calculated) than what is provided by the al- ready used sensors. We will show that using such \incor- rect" data may be bene¯cial for determining the position