A fuzzy qualitative model of a robot sensor, presented in this paper, is for locating 3-D objects, and the location information is used to guide the movements of an industrial robot to pick up the objects. The sensor consists of a rigid platform mounted on a flexible column. Each object to be located is held rigidly with respect to the platform. The static deflections of the column and natural frequencies of vibration of the dynamic system comprising the object, platform and column are measured and processed using a mathematical model of the system to determine the location of the object. In practice, the frequency measurements have low repeatability, which leads to inconsistent location information. Also, when the orientation is in the region 80°–90° relative to a reference axis of the sensor, the mathematical model becomes ill-conditioned. In this paper, a fuzzy qualitative model of the sensor is described. The fuzzy model is designed to yield the orientation in the region where the mathematical model is unusable. Different stages for constructing the fuzzy model are described. The on-line implementation of the model is outlined and the experimental results obtained are presented.