It is required lots of sensor systems to deal with general environmental uncertainties. However, in semi-structured outdoor environments, such as paved road, the environmental uncertainties are generally lower than that of the general environments. As a result, it can be dealt with a single Laser Range Finder by using the utilizations of curb-detection in the semi-structured environments. In this paper, we presented outdoor mobile robot localization scheme by exploiting the detected curb information. For localization, we used data of robot's odometry, DGPS, and compass was combined using Extended Kalman Filter. The data of extracted curb was also used. Using the data of curb, we can get more reliable localization results and suggest algorithm showing better localization performance than that of result from using only DGPS and compass in area in which DGPS's error is large.