Infrared motion sensing system for human-following robots

This paper presents an infrared motion sensing system for human-following robots. The proposed sensing system consists of two layers: a geometric sensing layer and a cooperative sensing layer. Through a modulation of the sensor field of view (FOV), the geometric sensing layer can generate bearing measurements of a moving human target from multiple perspectives. Through a fusion of bearing measurements from multiple perspectives, the cooperative sensing layer can localize the target and predict its moving direction for the robot to perform related actions. In this study, pyroelectric infrared (PIR) sensor arrays and Fresnel lenses are used for geometric sensing. The least square method is used for data fusion. The performance of the sensing system is evaluated through experiments of human localization and human following.