Goal reasoning for intelligent parameter adaptation in active sonar

Active sonar systems include a variety of parameters that can be dynamically tuned to improve system performance. In practice, however, parameters are typically set and fixed over long periods based on expected environmental conditions, prior performance, etc. Dynamic tuning of system parameters by sonar operators is impractical due to both the short time frame between adaptations and the complex relationship among the parameters, goals, and system performance. In intelligent active sonar, the system tunes parameters based on a set of goals and evaluation of how well those goals are being met. This approach significantly reduces the cognitive load on the operator, who can set high-level goals that are interpreted by the system and translated into low-level parameter adjustments. Goal-driven autonomy (GDA), for example, detects discrepancies between predictions and observations and generates context-specific explanations when significant discrepancies are observed. These explanations may indicate that new goals need to be introduced or existing goals de-activated. The explanation generation and goal formation aspects of GDA provide insight into how the intelligent system is reasoning about its actions and allow an operator to have direct input to the intelligent system when desired.