关键词: 移动机器人, 自定位, 序贯搜索, 直方图匹配, 核密度相关

Abstract: A twostep selflocalization method for mobile robots is presented based on laser information. After preprocessing of the laser scan map, straight lines are extracted by a sequential searching method during the first step and the angle histogram is constructed. Then the rotation angle of the robot can be computed by angle histogram matching. During the second step, the kernel densities of the laser data after histogram matching are estimated and the object function with translation parameter is constructed based on kernel density correlation. The translation vector is solved by the BFGS quasi Newton method. Experimental results demonstrate the effectiveness of this method.

Key words: selflocalization, sequential searching, histogram matching, kernel density correlation , mobile robot