• long endurance
• Extended Kalman Filter (EKF)
• energy recovery
• Autonomous Underwater Vehicles (AUVs)
• acoustic positioning systems
• long-term underwater navigation
• wave energy

This thesis addresses the problem of providing Autonomous Underwater Vehicles (AUVs) with long endurance capabilities from a twofold perspective.
On one hand, the problem of long-term navigation is addressed exploiting acoustic positioning systems: the considered scenario is that of a vehicle equipped with an Ultra-Short BaseLine (USBL) system as main aiding sensor, navigating within an acoustic sensors network acting as a Long BaseLine (LBL) infrastructure. Two different approaches are followed, in both of which acoustics is oriented to bound the typical drift of the inertial-based estimation error; the peculiarity of the proposed strategies lies in the fact the acoustic positioning information are embedded in a net- worked communication scheme. In the first one, the navigation algorithm is composed of two consecutive steps: at the beginning of the mission, the vehicle combines its GPS signal with the USBL measurements to localise the fixed transponders of the network. Then, the estimated positions of the beacons are used as LBL spots and the acoustic measurements are combined with the inertial ones to perform autonomous navigation. The fusion algorithm is formally cast into an optimal stochastic filtering problem and the navigation status of the vehicle (namely its position, velocity and accelerometer biases) is estimated by relying on classic Extended Kalman Filter structure. The second technique, instead, aims at executing the two phases of the previous method at the same time, exploiting acoustic measurements to simultaneously determine both the navigation status of the vehicle and the topology of the network, initially supposed unknown. The previous estimation algorithm is augmented with a dynamic database of the discovered LBL anchors, which are considered as features in the classic Simultaneous Localization And Mapping (SLAM) problems largely studied and developed in the visual navigation community: given this strong similarity, the presented approach is hence named Acoustic-based SLAM (A-SLAM). An extension of the A-SLAM algorithm to support the cooperative navigation of an heterogeneous team of AUVs is hence proposed, in which the vehicle equipped with the USBL acts as the team leader, while the other teammates are treated as mobile nodes of the acoustic network. Using the USBL measurements, the team leader can track the path followed by the mobile nodes, in addition to generate a map of the fixed nodes and self-localise within the network. Moreover, the navigation estimates are shared with the other teammates via acoustic communication to improve the localisation of the whole team. The performance of the proposed navigation algorithms are evaluated by post- processing of real data collected during several sea trials in the context of the THESAURUS (Italian acronym for “Techniques for Underwater Exploration and Archaeology through Swarms of Autonomous Vehicles”) project, devoted to the development of a team of cooperating underwater vehicles purposely designed for deep water archaeological search and documentation.
On the other hand, the development of a novel device capable of recharging the batteries of an AUV exploiting energy from the wave motion as potentially unlimited power supply is presented. The energy recovery device is the core of the WAVE (Wave-powered Autonomous Vehicle for marine Exploration) project, which aims to further develop the technology for autonomous, long-term marine missions with the design and the prototypal realisation of a novel, hybrid oceanographic glider/AUV with power recharging capabilities from environmental renewable sources, such as solar and wave energy. Within the project, an energy recovery system from sea waves motion was first considered and conceived as a module, namely the WAVE module, to be installed on-board a generic carrier vehicle. The carrier vehicle was then identified as the hybrid glider/AUV eFolaga+, in order to guide the development of the system towards a realistic experimental solution without losing the generality of the methodology itself. Moreover, the prototype vehicle can be equipped with additional, mission-oriented payloads. Three further modules were thus defined, each one enabling the execution of a particular task: environmental monitoring, seafloor exploration and acoustic communication. A high-level software is designed and developed in order to provide the user with payload management and mission control and supervision functionalities while guaranteeing to the vehicle a certain level of decision-making autonomy. The designed systems are installed on-board the control units of the corresponding modules and their effectiveness is assessed with extensive bench tests conducted in laboratory and preliminary sea trials. Although there is plenty of space for further optimisation of the system performance, the results emerged after the first experimentations of the prototype show the practical feasibility of the proposed approach.