A Voting-Based Sensor Fusion Approach for Human Presence Detection

Abstract

With advances in technologies, environments and human habitats are all set to become smarter. Such environments, which include smart homes, hospitals, campuses, etc., are oriented towards human comfort and safety. Detecting the presence of a human being in spaces within such environment forms a major challenge. Though there are sensors that can perform this task, they are not without limitations. Using information derived from either single or multiple sensors separately is not sufficient to distinguish human beings from other objects within the environment. Reliable detection of human presence can be achieved only by fusing information obtained from multiple sensors. In this paper, we describe an approach to human presence detection using a combination of PIR and ultrasonic sensors. Analysis is performed based on the received raw signals from the PIR as well as an analog ultrasonic sensor. A voting based approach has been used to classify signals obtained from human beings and non-human objects, thereby facilitating human presence detection. Results obtained from indoor experiments performed using this approach substantiate the viability of its use in real environments.