Robot self-localization is one of the most important issues to tackle in autonomous navigation.In order to meet the requirement of accurate localization

a self-localization algorithm based on multi-sensor information fusion is proposed.The kinematics model is made based on the analyzing kinematics architecture of mobile robot and the rigid restrictions;the measurement model of odometric and ultrasonic sensors are built based on the theory of sensors;the data provided by odometric and ultrasonic sensors are fused together by means of an extended Kalman filter（EKF） technique.Finally

the position of robot is reset by matched environment feature

and the position estimation of robot is given accurately.Experiment result proves that the proposed algorithm eliminates the cumulative errors of odometry obviously

and improves the localization precision efficiently.