Abstract: Considering the requirements for target localization of human-following robots,a hierarchical infrared motion sensing paradigm and its physical implementation using pyroelectric infrared(PIR) sensors are presented.From the perspective of sensing model,the proposed sensing paradigm is structured in two tiers.The bottom tier is the geometric sensing tier,and it is composed of geometric sensing units,which offer multi-view bearing measurements of a moving target within the field of view(FOV) of a robot.The upper tier is the cooperative sensing tier,and it coordinates the geometric sensing units at the bottom tier to localize the moving target.From the perspective of physical implementation,at the bottom tier,the geometric sensing units for measuring the bearing information of the moving target are constructed by combining PIR sensors and Fresncl lens arrays.At the upper tier,the least square optimization is used to fuse the multi-view bearing measurements to obtain the position information of the moving target.The experimental results are given to validate the proposed sensing method in the context of human-following with mobile robots.Compared with the multi-view optical vision techniques,the proposed method has irreplaceable advantages of high sensing efficiency,robustness to illumination and background changes, low-cost and low energy consumption.