移动机器人几何-拓扑混合地图的构建及自定位研究

基于墙角、房门和通路等高级环境特征的辨识与提取,依据几何和拓扑环境模型完成混合地图的构建,并根据混合地图的特点,提出在局部几何环境采用扩展卡尔曼滤波算法实现移动机器人的位姿跟踪,而在拓扑地图的节点位置则根据绑定的高级环境特征进行位姿再校正的混合定位方法.将该方法应用于实际移动机器人平台,所得结果证明了方法的有效性和实用性.

According to the practical result of geometric-topological map building, a hybrid localization approach is presented to complete both extended Kalman filter-based pose tracking in local geometric map and re-localization in topological node using the binding high-level features. The method is effective to avoid localization failure in the cause of unexpected collisions or unstructured dynamic environment. Experiment results implemented in the real mobile robot show the validity and practicability of the method.

中国科学院沈阳自动化研究所机器人学研究室基金项目(RL200204);;辽宁省高等学校学科拔尖人才资金项目(2003-54).