Arkadiusz Czuba
supervisor: Marek Nałęcz
Cognitive architectures are part of artificial intelligence. They have found an application in among other places, robotics, and autonomous driving. However, research related to the topic of integrating artificial intelligence architectures with cognitive radar is limited. In this work, cognitive abilities such as learning, perception, attention, and decision-making mechanisms were conformed to radar capabilities. As a result of that customization, a novel cognitive radar architecture has been proposed. This concept is introducing promising multimodal perception, visual attention mechanisms for resource allocation and target prioritization, memory-based cognitive models, and reinforcement learning for learning receiver-transmitter relation. Currently, the work is on the stage of implementation of the proposed system. The most similar cognitive capabilities to the proposed cognitive radar have Soar cognitive architecture. The validation implementation of AICR is based on the integration of Soar architecture with a radar system. Particular emphasis is placed on the development of human-like procedural and declarative memories adapted to the processing of radar knowledge. All of the aforementioned components combined with a memory-based cognitive control subsystem are looking promising to improve overall radar performance.