结合粗糙集理论,提出一种传感器数学模型客观确定方法.以移动机器人本体上应用最广泛的声纳传感器为例,分析声纳采集的大量客观数据之间的依赖关系,利用粗糙集理论的分类思想中正域、依赖度等概念,对按照一定规则离散化的声纳数据进行计算,获取传感器各个离散点的取样值,存入数据库中以备需要时查询.以PioneerⅡ机器人为实验平台,验证了方法的正确性和有效性.该方法可以推广到其他类型传感器建模中,不需要大量各领域专家的经验和知识,仅仅需要足够的采集数据即可建立不同的传感器模型,既避免了传感器建模的主观性又具有一定的通用性. Combining with rough set theory, a method to objectively determine the mathematical model of sensors is proposed. Taking the most widely used sonar sensor on the mobile robot as an example, the dependence of a large number of objective data collected by sonar is analyzed. The concept of the positive domain, the degree of dependence and other concepts, the discretization of sonar data in accordance with certain rules are calculated to obtain the sensor discrete points of the sample value, stored in the database to query if necessary.With Pioneer Ⅱ robot as a test platform to verify Method is correct and effective.This method can be extended to other types of sensor modeling, without a large number of experts in various fields of experience and knowledge, just enough data collected to establish different sensor models, both to avoid the sensor built Subjectivity of the mold has a certain versatility.