• acoustic localization
• acoustic sensors
• adaptive algorithm
• autonomous underwater vehicles
• AUV team
• cooperation
• cooperative algorithms
• data fusion
• environmental sampling mission
• event driven simulation
• exploration
• genetic algorithms
• geographical information system
• GIS
• harbour protection
• magnetic sensors
• mobile robots
• Monte Carlo
• one hydrophone localization
• optimization
• radial basis functions
• robotics
• security systems
• sensor models
• surveillance systems
• TDOA
• team of cooperating robots
• time difference of arrival
• underwater localization
• underwater robotics

This thesis focuses on the civilian protection, and suggests solutions based on the peculiar characteristics of such scenarios, specifically the severe limitations in testing an underwater detection system in the field and the limited economic resources available.

The first part of this work focuses directly on the main security problem, describing a new approach, based on simulating tools, to analyze and verify harbours protection levels. Traditionally based on trial-and-error methodologies, building an underwater anti-intrusion system is a complex operation, starting from choosing the right sensors to be placed in the specific scenario to configuration deployments into the water. Using a simulating tool can be an efficient way to get better information on the anti-intrusion behaviour before dipping any sensor into the water and then with a very limited influence on the normal port activities. More specifically this could even be the only solution for civilian ports protection. The simulator tool is based on a GIS architecture, acoustic and magnetic sensor models and an event driven simulator to simulate the sensors performance in their relation with the specific sea environment and intruder classes (diver, ships, etc), providing operational and ready-for-action data in order to build or modify an anti-intrusion system.

The second part describes four low-cost monitoring systems for marine areas and the major problems connected with the underwater localization. All the systems, based on passive acoustic hydrophones, can be used to protect and monitor non-critical zones such as fishing farming installations, archaeological sites or protected marine areas, and in general in situation in which access may be always or temporary forbidden to unauthorized personnel.
The first system shown is based on only one hydrophone, using a very simple acoustic propagation model, to track and localize the threat. Two and three hydrophones systems are then described making the installation more complex and expensive but effectively increasing the localization performance and reliability.
Finally, the discussion goes further using an autonomous underwater vehicle with very limited computational and signal acquisition capabilities to localize a fixed underwater source of sound. In this part the problem is not in the localization itself, for the vehicle can carry up to four hydrophones, a number that is sufficient to localize exactly a sound source in a three dimensional scenario, while the challenge is in the very limited vehicle acquisition resources which make useless all the traditional localization algorithms based on reliable signal delays among the hydrophones.

Besides, while the security problem has to be investigated directly, underlining systems performance with respect to specific threats, on the other side, it involves even several side tasks that have to be accomplished to make the general one completed. Protecting marine areas means having the best knowledge of the areas we are interested in, and then providing a large amount of data both from historical databases and from new experiments in order to get up-to-date geographic and environmental information.
The third part of the thesis is in fact related to a very important but side field in the whole security aspect and it concerns with marine areas exploration through a platoon of autonomous underwater vehicles (AUVs) performing point-wise measurements or observations.
The proposed algorithms, based on a fuzzy like approach or on the use of radial basis functions (RBFs), allow to adapt vehicles exploration to the specific conditions the agents are into, without any need of an a-priori knowledge of the mission area, incrementally controlling the approximation error of the overall environmental quantity as the mission goes on and new scattered data are provided. The algorithms can be applied to more general cases in which point-wise measurements are required and mobile sensors networks are used, even though, in this thesis, they are especially aimed to environmental monitoring and exploration of oceanic regions.