The earth's surface comprises 70% water, with a market value of coastal resources estimated around 3 Trillion USD for year. Coastal and marine ecosystems play a vital role for life on Earth, and they are central to a vast range of industries and people activities. Therefore, disaster detection, cost surveillance, as well as environmental data gathering are key activities at sea for countries and authorities.
However, monitoring activities are hampered by the vastness of the area, and by the inherently hostile and harsh environment. Current systems fail short in providing an accurate view of present situations due to the impossibility of covering the whole area, the technology limitations in gathering detailed data, and limited personnel and equipment.
Novel solutions exploit satellite technologies to gather images and monitor the coastal area, providing however low accuracy (i.e., low-resolution data with huge delays). Differently from these solutions, this project proposes squads of drones (DroNets) to autonomously coordinate in maritime environments. While DroNets are increasingly used for monitoring and patrolling onshore fields, we aim at exploiting them to support existing coastal and port surveillance systems. Squads of Drones are scalable, fast, and resilient, proving the perfect tool in maritime monitoring systems.
Our approach is innovative because it provides a completely automated system to monitor coast and sea-port areas. We design algorithms that autonomously guide drone trajectories for the first time into a maritime environment, which is inherently hostile and harsh to monitor. Moreover, the use of Reinforcement Learning techniques enables to dynamically adapt drone trajectories, and refined monitoring tasks, upon the current weather conditions, and local findings.
This is innovative concerning the current state of the art in many aspects:
1) Current maritime monitoring systems rely on satellite images and human personnel, which are unable to have enough high-resolution and updated data to achieve a complete situation awareness. Instead, our approach aims at carrying a scalable swarm of drones, to automatically monitoring and detect possible events.
2) Current drone systems are mainly developed for onshore missions, while we propose an autonomous network of drones able to work in the hostile and hard maritime environment