基于FastSLAM的RoboCup家庭组机器人自定位方法研究

Study on Self-localization System of RoboCup@Home Robot Based on FastSLAM

FastSLAM算法为机器人定位与地图构建开辟了新的研究方向,研究该理论在RoboCup家庭组机器人自定位系统上的应用。从贝叶斯滤波理论,Rao-Blackwellised解耦,粒子滤波器原理等方面论述了FastSLAM的关键技术及基本理论。针对家庭组机器人非结构化场景的特殊情况,提出基于FastSLAM的定位策略。仿真结果表明算法的有效性和鲁棒性,能够为家庭机器人在非结构化场景中的定位问题提供有效地解决方案。

FastSLAM theory has opened a new research direction for robot localization and mapping, and its application on Self-Localization System of RoboCup@Home Robot was studied. The key technology and basic theory of FastSLAM were covered from aspects of Bayes filter theory, Rao-Blackwellised factorization and particle filter. A localization method based FastSLAM was proposed, specifically to the unstructured scenes for RoboCup@Home Robot. Simulation results prove the effectiveness and robustness of the method which can provide effective solution for the self-localization problem of RoboCup @Home Robot under unstructured scenes.