Combining Technologies for Aiding Search Missions with Drones

This paper presents the combination of technologies in an application for aiding search missions. Unmanned Aerial Vehicle (UAV), wireless communications, and image processing and classification technologies are combined for generating warnings to an operator in a search mission. Thus, the UAV would be used to cover a specific ground area communicating images the system should be able to classify. In case a positive image is classified (i.e., one with the object/s being searched for), an operator is warned so that the final decision is made by an expert. In this way, the mostly bored task of looking at images without object being looked for is avoided, and the expert attention could be focused on images with a high probability of having the object/s being looked for. A cloud service is used for image classification (IBM Watson) along with a previously proposed UAV system architecture and a low-cost drone SDK. A proof of concept is described, giving implementation details, and preliminary results. Obtained results show that the general system is a viable solution and set a starting point for enhancing several subsystems for better performance.