SWIFTERS: A Multi-UAV Platform for Disaster Management Maria Terzi, Antreas Anastasiou, Panayiotis Kolios, Christos Panayiotou, Theocharis Theocharides Abstract-In this paper we present SWIFTERS, a technology platform for utilising multiple UAVs in disaster management. The platform aims to provide fast situational awarene;"s and improve the effectiveness of disaster management operations by exploiting UAV deployment opportunities. The platform allows flying UAVs in a click-and-go fashion, scanning an area using multiple collaborating UAVs while receiving live video feed from the camera of the UAVs' to detect people, vehicles and carry out visual disaster assessment. This paper provides evidence of the reliability and scalability of the SWIFTERS platform in a real-life experiment with four UAVs. Index Terms-Unmanned aerial vehicles, Multi-agent systems, Emergency services, Software systems. I. I NTROD UCTION Unmanned Aerial Vehicles or simply UAVs, have shown their potential in multiple areas of our lives including tackling some of the most challenging tasks to human beings such as natural disasters [I], [2], [3]. Although the natural disasters cannot be avoided, their effects could be significantly mitigated through a comprehensive and highly efficient disaster manage- ment system. Time is the most essential factor in successful disaster management and hence disaster response must be designed and carried out swiftly based on accurate assessments of the situation. Conventional disaster management systems typically rely on ground emergency respon se efforts to carry out disaster assessments. However, such approaches are resource hogging and greatly affect the disaster assessment time. Additionally, following a disaster event, access on the ground may be severely constrained; making it impossible for the first re- sponsders to effectively carry out their tasks, as indicated in Lettieri et al. [4]. UAVs have the potential to enhance and optimise all phases of disaster management including disaster assessment [5]. At the pre-disaster stage, UAVs can be used to accurately predict the outbreak and assist in disaster prevention activities with only a fraction of the resources needed by conventional operations [6]. During the disaster, UAVs can provide high- resolution, real-time views of the most inaccessible locations which can then be used to produce high-quality maps and aid rescuers in accurately assessing the situation, provide relief and conduct rescue [7]. In the post-disaster stage, UAVs can be used to map the affected areas and aid in the reconstruction of the damaged infrastructure [8]. First responder units (such as police departments, fire brigades and civil protection) that have integrated UAV plat- forms in their operations rely merely on their manual control. The authors are with the KIDS Research and Innovation Centre of Excellenc e (KIDS CoE) and the Department of Electrical and Computer Engineering, University of Cyprus, Nicosia, 1678, Cyprus: e -m a i l : {t e r zi. m ari a, anas tas iou .ant reas , pk o110s chr istosp ,tth eocharides }@ucy .ac .cy This is achieved by either using the remote controller of the UAV or via a mobile phone application that provides generic UAV control features [7]. However, manually oper- ating UAVs requires a sufficient number of trained pilots and accompanied personnel (visual observers, spotters, technical support) imposing a great challenge in the available human resource pool. Also, it increases the stress caused to the team members (as indicated in [9]) and the operation is more prone to errors especially when undertaking multi-UAV tasks; such as searching of an area using multiple UAVs to minimise the disaster assessment time [10]. Multi-UAV platforms enable the use of a fleet of aerial units to conduct numerous activities in parallel and handle flying patterns to maximise performance. In addition, the operator of the system maintains a global view at all times (hence improving safety) , and receives live feed from the on- board sensors (e.g. cameras) for acquiring real-time situational awareness. Currently, the available multi-UAV platforms are restricted in the type of UAVs they support and the features that they provide especially for emergency response operations. For instance, the offline use of the system (i.e., without Internet connection) is a key feature in emergency response missions that has not been looked at to date. Multi-UAV operations have also not been introduced to existing solutions and is one of the main features that our proposed system aims to address. Hence , the main contribution of this paper is the design , development and implementation of a multi-UAV platform dedicated to assisting response units throughout the life cycle of disaster management. The platform enables real-time mon- itoring (using the onboard sensors) , live video transmission from all connected UAVs, video analytics (object detection done in real time) and mission planning tools using state-of- the-art multi-agent algorithms, taking into account the chal- lenges posed by the UAVs and the operational environment. The platform is designed to enable integration with modules developed in any programming language and framework. It can support any type of programmable UAV through a dedicated interface, includes a database for storing persistent data and can run completely offline from a single or more computing devices. To demonstrate the applicability of the proposed solution, a real-life experiment was carried out with accompanied results to evaluate the performance of the proposed software platform. The results presented hereafter provide evidence of the reliability and the scalability of the proposed SWIFTERS platform. The rest of the paper is split into the following sections. Related work is included in Section II while Section III elaborates on the proposed framework and presents in-detail the system architecture and discusses some of the most important components. Section IV presents the methodology 978-1-7281-4920-2119/$31.00 ©2019 1EEE Authorized licensed use limited to: University of Cyprus. Downloaded on June 03,2020 at 13:52:03 UTC from IEEE Xplore. Restrictions apply.