公开(公告)号：KR1020140009737A

公开(公告)日：2014-01-23

申请号：KR1020120076299

申请日：2012-07-12

Applicant: 한국과학기술원

Inventor： 권인소 ,  최동걸

IPC: B25J13/08 ,  G05D1/02

Abstract: The present invention provides a position recognition method of hybrid map based robots to effectively respond to situations such as kidnap problems in an environment without GPS and to accurately recognize current positions, including processes of acquiring information on surrounding environments using a laser sensor and a camera; determining laser key frames from the laser information acquired through the laser sensor to generate a metric map; determining image key frames capable of expressing each feature from the image information acquired through a camera to generate a topological map; connecting the laser key frames and the image key frames by determining weights according to the sensor characteristics along with determination of integrated key frames; and connecting a local metric map prepared by using ICP algorithm between integrated frames with each laser and image key frame according to differences in a distance from the integrated key frames, thereby recognizing the current position. [Reference numerals] (S12) Laser information acquisition; (S14) Image information acquisition; (S20) Metric map generation; (S30) Topological map generation; (S40) Integrated key frame determination; (S50) Current position recognition

Abstract translation: 本发明提供了一种基于混合地图的机器人的位置识别方法，以有效地响应在没有GPS的环境中的绑架问题等情况，并准确地识别当前位置，包括使用激光传感器和相机获取关于周围环境的信息的处理; 从通过激光传感器获取的激光信息确定激光关键帧以产生度量图; 确定能够从通过照相机获取的图像信息中表达每个特征的图像关键帧，以生成拓扑图; 通过根据传感器特征确定权重以及确定集成关键帧来连接激光关键帧和图像关键帧; 并且根据与集成关键帧的距离的差异，将通过使用ICP算法准备的局部度量图与每个激光和图像关键帧连接在一起，从而识别当前位置。 （附图标记）（S12）激光信息获取; （S14）图像信息采集; （S20）公制图生成; （S30）拓扑图生成; （S40）综合关键帧确定; （S50）当前位置识别