The Potential Role of Digital Twins for UAS/Ship Interface Artificial Intelligence (AI) Systems

The traditional perception of a digital twin in shipbuilding is as an initial CAD design to support the physical build process. More recently it has become a system in which data from the operational ship is maintained in approximately real time. One of the most challenging tasks in the deployment of basic ship-based VTOL Unmanned Air System (UAS) is the launch and recovery and deck handling operational performance on smaller ship platforms like Corvettes, Frigates and Destroyers. This is owing to the random nature of seaway created deck motion coupled with ship structure disturbed air wake patterns. For most of the traditional VTOL UASs, the precise measurement of these environmental data is essential for air vehicle safe launch and recovery events. Placing this real-time digital “shadow” within a realistic UAS x ship environmental simulation enables its use for planning and scenario testing, and for predictive operational mission planning. As Artificial Intelligence (AI) systems begin to be incorporated into ships, digital twins provide a new opportunity. A high-quality digital twin embedded in a realistic environmental simulation presents the possibility of generating large volumes of UAS x ship interface training data for the AI systems, without the need for expensive sea trials. An example is presented on of how this might work and discuss the benefits and challenges of such a system. The class of technology represented by this AI and QPP approach is to measure remote sea surface profiles to predict the future wave forces acting upon a vessel. An example being closely followed achieves this by using a neural network (artificial intelligent) to predict deck motions based on photogrammetric remote sea surface measurements. The photogrammetric sea surface measurement system calculates the sea surface several hundred meters in advance of the ship. The Arial University/IAI computational methodology shows good correlation between observed forecasted and real ship motion recordings measuring up to 4 seconds for roll and pitch alone. QPP’s approach functions in all-weather conditions. The photogrammetric method is not considered operationally all-weather. Nonetheless, the Arial University promises to be functional if it can select a more robust sea surface measurement system coupled with a neural network matrix composed of all 6-degrees-of-freedom.