Sensor Fusion-based Cooperative Human Detection Scheme using MAVs in Indoor Environment

Human detection has been a challenging area of research with its main applications such as rescue, surveillance, and defense. Many researchers have endeavored to improve accuracy in human detection using multimodal sensor fusion, which analyzes and correlates data from multiple sensors. Recently, Micro Aerial Vehicles (MAVs) have been adopted in safety and security domains due to their high mobility, but they could not be equipped with multiple sensors under strict weight limits. Indeed, a MAV with only one sensor has difficulty in detecting human with high accuracy in dynamic environment. In this paper, we propose a novel cooperative human detection scheme based on multimodal sensor fusion. The proposed scheme incorporates multiple modalities including appearance, motion, and voice. We implement a prototype system consisting of multiple MAVs enabling real-time processing on resource-constrained hardware. To verify the feasibility of the proposed scheme, we perform cooperative human detection experiments with multiple MAVs in indoor environment.