Accurate self-localization of mobile robots based on vision sensors

Xiaoming Dong

Department of Information and TechnologyAnqingTeachers college, 128 Linghu SouthRoad, Anqing, Anhui province,China

Robot localization is a challenging problem in indoor environment since no GPS information is available. In this paper, an algorithm was proposed for accurate localization, which designed a delicate way to extract the feature points at first, then the position of the  robot was determined using the relation of the features in different images, finally, the Kalman filer was designed to decrease the error caused by robot’s moving. Experiments showed the accuracy and robustness of this algorithm.