The guide and control of Autonomous Underwater Vehicles (AUV) have been subjected to a great interest from industry and academia in the past decades due to their scientific challenges and to an increased number of commercial and military applications including underwater resources exploration, oceanographic mapping, undersea wreckage salvage, cable laying and inspection, geographical survey, coastal and offshore structure inspection, harbor security inspection, mining and mining countermeasures (Fossen, 2002).

As an example of adverse conditions where the autonomy of AUVs’ control actions is of special importance is when they experience faults or failures while executing underwater tasks. Typically, thrusters are known to be one of the most likely sources of faults. In certain cases, the existence of a thruster fault results in the termination of the ongoing mission. The implication of these faults could be very expensive and time consuming. Therefore, research and development to produce a fault-tolerant system for the AUVs capable to reconfigure the control action under faults has gained much attention over the past years.
Also at the guidance control level, research efforts are ongoing to provide vehicle trajectory generation in a completely autonomous way. Traditional methods for guidance of AUVs employ sonars, magnetic sensors, acoustic transponders and optical sensors. Optical imaging sensors (e.g. TV camera), on the other hand, are most effective at shorter distances, where the effects of forward and back-scattering are less dominant. Optical imagers are thus the systems of choice for applications that require high image resolution at close range, such as station keeping, control of manipulators, object identification, cable following or salvage and retrieval. Computer vision will play an important role in achieving underwater autonomous systems. However, visual guidance of AUV is currently accomplished by relaying the video data to topside operators via high bandwidth data links such as optical fibers or hard cables. Merely replacing the operators with powerful topside computers will not be enough. To achieve the highest degree of freedom and truly enjoy the full advantages of an autonomous vehicle, the information processing must be accomplished aboard the AUV itself (Nguyen et al., 1989).

The proposed research activity will focus on the study and testing of new strategies for the control and driving of autonomous underwater vehicles (Autonomous Underwater Vehicles - AUV) for the visual inspection of submerged infrastructures of industrial interest.

A second research topic will concern the study and experimentation of methods of autonomous localization of the vehicle with respect to the submarine environment and the generation of trajectories aimed at the optimal achievement of the mission objectives. To this end, it is essential for the AUV to have the ability to create a visual map of the submarine environment, locate the vehicle with respect to this map (Simultaneous Localization and Mapping - SLAM) and guide it by following certain patterns on the map, for example submarine piping by orientating the vehicle parallel to it. The traditional methods for localization and visual guidance of AUV employ optical sensors, possibly supplemented by ultrasound acoustic sensors.

A third line of activity, more to character industrial and experimental that will take place prevalently in the company, will cover the study of the main techniques and best-practice of 2D-3D reconstruction of marine environments and the main technologies used for underwater photogrammetry. The study will include the planning of the optical / acoustic inspection mission of submerged assets, the collection of their images, their processing and possible fusion, their geo-referencing, also through acoustic communication of GPS data with the support boat.