© 2019 Association for Computing Machinery. Thanks to technological advances in information and communications, Unmanned Aerial Vehicles (UAVs) (aka drones) technology has become one of the most important service delivery inventions these days. Equipped with sensors and cameras, these technologies can perform much on-demand critical application ranging from military and environmental to rescue operations. These UAVs devices, sometimes tiny, that can replace human and manned aircraft in several tasks, have been utilized to perform various types of services and applications more efficiently. However, the deployment side of such emerging technology is still facing many issues and challenges. Using multi-drones or swarm of drones, together to accomplish one operation is one of these recent challenges. Such a system requires a high level of delicacy and cooperation to achieve the required autonomy and reduce human interaction as possible. Communication is one of the biggest challenges in these systems, since the devices should keep exchanging various types of messages with different Quality of Service (QoS) requirements. In this paper, a collaborative autonomous system of swarm drones based on deep learning has been proposed and a testbed of cooperative UAVs' mission to validate the performance of the dedicated QoS communication system using Paparazzi drones The proposed system is aware of the drones' requirements in term of QoS and able to meet with their dynamic demands.