Sistema de Preservação dos Oceanos Baseado em Gémeos Digitais e Veículos Heterogéneos não Tripulados

The Portuguese Navy is currently responsible for receiving alerts from CleanSeaNet and taking necessary actions to respond to ocean spills [1], including verification to exclude false positives. This requires significant resources, such as manpower, assets, and costs, from a small country with limited funding and an extensive coastline, as well as one of the largest Exclusive Economic Zones in the world. By using a fleet of heterogeneous UVs (aerial, surface and underwater), it is possible to mitigate the delayed alerts of CleanSeaNet. Additionally, machine learning algorithms can be employed to analyze data from European Maritime Safety Agency (EMSA), weather reports, sea conditions, and in-loco monitoring parameters to generate valuable information and insights that can be used to optimize resources for both future monitoring and response activities, while minimising response time. The challenges of a mechanism for the surveillance of such a large Exclusive Economic Zone are various, diverse and its surveillance depends on favorable weather conditions. Besides that, there is the surface and the submarine zones and the areas near the coast and far from the coast. Surveillance through human resources on land is not always feasible due to weather and water conditions, can be expensive and have a short-range coverage. Heavy aerial and surface surveillance mechanisms, such as airplanes, helicopters, and boats, also come with high operating and maintenance costs, as well as the need for human resources. Additionally, these approaches are seldom environmentally friendly. One possible solution is to use UVs. However, using individual vehicles cannot respond in real-time to all areas, requiring more autonomy and time to scan large areas and representing a single critical point of failure. If a vehicle fails, the entire system is compromised. Sea pollution projects and systems often focus on the during and after ocean spills stages, which are often too late. However, we think these challenges are not fully addressed. In a country with limited resources like ours, it is advantageous to allocate the scarce resources effectively during and after disasters, as well as to act preventively by identifying possible early signs of problems. This project proposes an innovative approach to predict, prevent, and detect ocean spills. The approach utilizes technologies such as UVs carrying infrared, RGB, and multispectral cameras, as well as AI, to detect early signs of imminent spills, whether deliberate or accidental. The aim is to create a predictive, low-complexity, low-cost, ever-learning, scalable, easy-to-use, and effective mechanism for detecting spills early. The employed innovative technologies include DTs for the UVs, automatic vehicle guidance algorithms, forecasting and prediction of spills using AI mechanisms, and forensic analysis. To the best of our knowledge, no works have been proposed or developed on the prevention, prediction, and detection of ocean spills using DTs, integrated with external national and international data sources (weather conditions, EMSA, etc.), and combining the cooperation of heterogeneous UVs, ML techniques, and DTs. The project team, supported by the expertise of relevant partner institutions and national authorities, has the necessary background, knowledge and experience to achieve the objectives and expected results, allowing the project to progress from a TRL 2 to a TRL 6 prototype.