Abstract: For the self-localization problem of the indoor mobile robots，a self-localization method of indoor mobile robots based on artificial landmarks and stereo vision is proposed.First，a landmark model is designed to give position information；second，color segmentation，invariance of cross-ratio and the self-adaptive window are used to detect the artificial landmark，so the position information is got；finally，in order to locate the mobile robot accurately，the binocular stereo vision is used by combining with the position information of artificial landmark.The experimental results show that the proposed landmark model has strong robustness to the light and the camera position，and the self-localization accuracy can meet the requirement of indoor mobile robot.

Key words: vision localization, artificial landmark, stereo vision, mobile robots

摘要： 针对室内移动机器人的自定位问题，提出一种基于人工路标和双目视觉的室内移动机器人自定位方法。首先设计了一种可扩展的彩色人工路标，并给出路标的编码方法；然后利用色彩空间变换，直线交比不变性以及自适应窗口实现路标检测与识别；最后在分析双目立体视觉模型的基础上建立起基于路标的双目立体视觉定位模型，实现移动机器人的准确定位。实验结果表明，路标对光照和视觉传感器的采集位置具有较强的鲁棒性，定位精度能够满足室内移动机器人的定位要求。

关键词: 视觉定位, 人工路标, 立体视觉, 移动机器人