A SLAM method based on memory database of the features

A method for simultaneous localization and map based on scan matching is presented according to memory database of the features. the steady geometric features are extracted as the natural landmarks according to advanced least square fitting, and the memory database of the features is created and updated for scan matching better. Therefore, it is also valid to dynamic environment. Meanwhile, the belief is imported to reduce the result errors caused by the uncertainty of the sensors. the experiments on mobile robot proved the validity and accuracy.