哈尔滨工程大学学报2017，Vol.38Issue(12)：1906-1914,9.DOI:10.11990/jheu.201610060
基于人工离线特征库的室内机器人双目定位
Binocular localization of indoor robot based on artificial offline feature-database
摘要
Abstract
In this paper, a binocular vision localization method adopted in a comprehensive environment with dense feature points is proposed in order to address the problem of mobile robot localization in a complicated indoor envi-ronment.To overcome the computational complexity difficulties at the moment when more feature points with a weak ability to data association existing, which establishes a feature-database by an artificial offline algorithm, was de-signed on the basis of the extended Kalman filter-simultaneous localization and mapping ( EKF-SLAM) method. Improvement measures for the sparse control mechanism and salient feature point selection were introduced in order to guarantee the distributed-uniformity of feature points and controllable feature-database.The test result revealed that this algorithm can realize real time positioning with high reliability and accuracy.Moreover, in this method, a periodically updated strategy was employed in order to achieve adaptation to environmental changes.Therefore, long-term effective localization could be realized.关键词
移动机器人/室内机器人定位/双目视觉/稳健特征点/离线建库/稀疏性策略/EKF-SLAMKey words
mobile robot/indoor robot localization/binocular vision/salient feature point/offline establishing fea-ture-database/sparse strategy/extended Kalman filter-simultaneous localization and mapping分类
信息技术与安全科学引用本文复制引用
张智,张磊,苏丽,郭文县..基于人工离线特征库的室内机器人双目定位[J].哈尔滨工程大学学报,2017,38(12):1906-1914,9.基金项目
国家自然科学基金项目(51409053) （51409053）
黑龙江省自然基金项目(E201414). （E201414）
