Application of machine learning and artificial intelligence techniques to improve autonomy in maritime surveillance radar systems.

Modern radar systems are extremely complex devices with multiple modes of operation, each tailored to a specific suite of tasks. Consequently skilled operators are required to monitor the output, select the appropriate modes and interpret the data to support current operational objectives. As objectives change from mission to mission, the relative importance of each piece of collected data changes also. For example, detection of a piece of debris on the ocean surface is not important when patrolling shipping lanes in search of pirate activity, but is extremely important when searching for wreckage of a missing aircraft. This distinction is obvious to a human operator, but not to an autonomous system. The aim here is therefore to investigate methods for inserting autonomy into existing maritime radar systems for application on autonomous platforms.

In this project we will use techniques from artificial intelligence and machine learning to provide the radar system with a degree of goal-directed autonomy. This will be accomplished via simulation using a complex, state-of-the-art multi-resolution multi-agent simulation engine (MAVERIC) developed at the University of Glasgow. Existing radar models will be refined and new high-fidelity models created with input from the industrial partner on this project, SELEX Electronic Systems, Edinburgh. Machine learning and AI techniques include, but are not limited to, finite-state machines, Bayesian classification and neural networks all informed with heuristics supplied by MAVERIC and SELEX.