In the project we integrate advanced sensors, such as multi-wavelength fluorometers and underwater laser cameras, to state-of-the-art robots and develop machine-learning algorithm for automatic identification of marine features and marine organisms. Monitoring and management of marine environments and infrastructures is a challenging and costly activity due to the considerable heterogeneity of the oceans, which display large changes in water conditions (e.g. current, visibility), water quality (e.g., nutrient loads, carbon content) and marine life, at all scales from few meters to thousand of kilometers. There is a need to achieve cost effective solutions in infrastructure inspections as well as water sampling, analyses and observations performed directly in the ocean (in-situ) and at temporal and spatial scales relevant for the processes investigated. This requires solving specific challenges since: a) in-situ sensors and analytical probes are still limited; b) collection of long-time series is problematic; c) infrequent sampling cannot capture transients and episodic events.
SENTINEL will focus on application and demonstration of cognitive robotic systems in underwater environments and in close collaboration with large Danish Maritime Industry Partners.
The system is composed by four elements: 1) energy-efficient thruster driven underwater vehicle with high maneuverability and control; 2) sensors for real-time environmental data including water quality properties and visual detection; 3) cognitive system composed by machine learning algorithms converting sensors information into features identification; 4) sampling device to collect water samples to locations identified by the cognitive system.