Using the assumption of the horizontal plane

a method of fast localization for mobile robot is presented. This method addresses an Obstacle-Ground-Corner extraction algorithm combined with the Harris-Laplace operator

fast mean shift and stereo vision

and then uses the obstacle ground corners as landmarks. The adaptive particle filtering method is used to fusion the landmarks depth and position to achieve an efficient and more accurate fast localization

the computational complexity of localization algorithm is greatly reduces due to the use of a small number of landmarks. The results of experiment illustrated the effectiveness of this method.