CERTH is the Technical manager of NERITES project, supporting the PC to monitor any potential technical risks and mitigate relevant innovation and technical bottlenecks. Moreover, CERTH is the leading beneficiary for developing the autonomous capabilities of the controllable degrees of freedom available from the underwater torpedo-shaped UAV (e.g., thrusters, gliding wings, etc.), as well as its sampling and metering end-effectors (e.g., docked robotic arm for underwater sampling). On this matter, CERTH is going to capitalize on deep reinforcement learning approaches to train robust agents for UAV path-planning, maneuvering and sampling end-effectors control  Moreover, CERTH will develop a comprehensive toolbox for assessing the preservation state of Underwater Cultural Heritage (UCH) using advanced data-driven modeling techniques. This toolbox will leverage deep learning methods, including variational autoencoders and convolutional LSTMs, to ensure high data quality and anomaly tolerance, supporting critical decision-making for UCH preservation.