The ISC is part of the Johns Hopkins Applied Physics Laboratory and will follow all current policies. Please visit the JHU/APL page for more information on the Lab's visitor guidance.


Search-Based Testing Methods for Evaluating the Resilience of Autonomous Unmanned Underwater Vehicles


The resilience of an unmanned underwater vehicle (UUV) can be defined as the vehicle’s ability to reliably perform its mission across a wide range of changing and uncertain environments. Resilience is critical when operating UUVs where sensor uncertainty, environmental conditions, and stochastic decision-making all contribute to significant variations in performance. A challenge in quantifying the resilience of an autonomous system is the identification of the performance boundaries—critical locations in the testing space where a small change in the environment can cause a large change (i.e., failure) in an autonomous decision-making system. This article outlines a methodology for characterizing the performance boundaries of an autonomous decision-making system in the presence of stochastic effects and uncertain vehicle performance. This approach introduces a method for hierarchically scoring the autonomous decision-making of these systems, allowing the test engineer to quantitatively bound the performance prior to UUV deployment. When using this scoring approach, engineers apply a set of novel subclustering methods, allowing them to identify stable performance boundaries in stochastic systems. The result is a process that effectively measures the resilience of an autonomous decision-making system on UUVs.