Abstract: A probability model for a laser scan matching to build global map for mobile robot.Most laser range scan matching algorithm are based on the features（points or lines） correspondence relation have to be established，such as Iterated Closest Point（ICP） algorithm.Using normal distribution transform probability model to represent a laser range scan.Not need features corresponded，but transform the discrete scan date to the piecewise continuous and differentiable probability density function，and using Newton’s algorithm to optimize scan matching，so can build the 2D global map.Experimental results demonstrate that the method is capable to map unmodified indoor environments reliable and in real time.

Key words: scan matching, probability model, normal distribution transform, 2D map building

摘要： 提出一种激光扫描数据匹配的概率模型，用于移动机器人的全局地图建立。大多数的激光扫描匹配算法都需要利用特征（点或线）来建立对应关系，例如ICP算法。利用正态分布转换概率模型来表示激光扫描，不需利用特征对应，而是通过将扫描得到的离散数据点转换成分段连续可微的概率密度函数，并利用牛顿优化算法来进行扫描匹配，从而建立2维全局地图。实验结果表明，该方法可有效地实时完成室内环境下的2D全局地图建立。

关键词: 扫描匹配, 概率模型, 正态分布变换, 2维地图建立