Goal Driven Autonomy as a Model for Reasoning in Cognitive Active Sonar

摘要

While active sonar systems include a variety of parameters that can be tuned to improve performance when addressing search and track goals, such parameters remain largely fixed, in part because tuning individual parameters imposes an impractical cognitive load for sonar operators. In this paper, we propose adopting a goal reasoning architecture to develop an intelligent active sonar system that can dynamically tune parameters to meet system goals. Goal reasoning architectures support systems that reason about, prioritize, and generate goals based on information obtained in real time. Using the goal-driven autonomy architecture as a framework, we have developed a model for an active sonar system that can tune parameters at both the transmitter and receiver to improve its ability to perform search and track tasks. We provide a proof-of-concept example in which an intelligent sonar system whose goal is to track a maneuvering target uses observed Doppler from the target to decide what waveform to transmit. Simulation results show that intelligent waveform selection yields improvement with respect to both target location estimation and maintaining the target track, providing evidence to support the goal reasoning model for intelligent sonar. Next steps include developing mechanisms to arbitrate among simultaneous system goals.