Sensor fusion-based human tracking using particle filter and data mapping analysis in in/outdoor environment

This paper proposes a method to track an object for a person-following mobile robot, which can complement disadvantages of various sensors. For human-robot interaction, a mobile robot should maintain a distance between the person and itself. Maintaining this distance is divided into two parts: (1) the object tracking and (2) the person-following. The object tracking consists of a particle filter and online learning using shaped features, which are extracted from an image. However, a monocular camera may fail to track a person because of the narrow field-of-view and influence of illumination changes, therefore, the laser scanner has been used together with the camera. After getting the geometric relationship between the differently oriented sensors, the proposed method will successfully track a person. The experimental results show a 93.3% success and robustness in both an `in' and `outdoor' environment DB.