Automated acoustic interrogation of seabeds for identification and detection of abandoned munitions requires that system algorithms be robust enough for a wide variety of environmental and target conditions. Empirical tests are a crucial part of the verification of robustness, but even more basic is a physically justified design of the algorithm and feature selection. The objective of this SERDP Exploratory Development (SEED) project was to develop a finite element (FE) approach to aid in incorporating the target physics. This research provides a foundation for further development of individual physics component modeling.
Automated acoustic interrogation of seabeds for identification of abandoned munitions will enable the application of mature sensor technology to the vast data throughput necessary in practical application. However, unmanned systems in particular require high confidence in system robustness, and this has not been achieved to date given the limited physical motivation even for algorithms that currently perform well in limited tests. FE modeling of munitions has allowed high-fidelity prediction of scattered acoustic returns from sources too complex to analyze using empirical models. The researchers’ laboratory has been focused on developing specific FE techniques not just to predict responses but to isolate the effects of individual physical mechanisms contributing to the response. Because complex target responses are broken down into individual components in this approach, it is more straightforward to understand the environmental effects on such components than on the complete return. In this project, the researchers investigated the effects of burial state (amount and material) on these individual components and used the results to predict the complete response and provide a physically justifiable assessment of the robustness of features that are being used or proposed for automated munitions identification.
This SEED effort demonstrated proof-of-concept for the value of isolating the contribution of various aspects of the physics contributing to a complex elastic response of a target. In this work, particular regions of the target were loaded as a method for better understanding the full-physics response and the way in which this response will vary with environment and target conditions.
Two basic approaches were explored: coupling-angle-focused and depth-focused. The former was motivated by ray-theory research that recognizes certain phase-matching conditions are necessary for certain types of fluid-structure coupling, and suggests a direction for isolation and identification of the corresponding scattered-wave features. Isolating features by the depth of the target surface on which they couple was also looked at; this is motivated by the awareness that in the field the targets may be partially buried and thus the ability to characterize the features by depth will be directly relevant. For comparison, the full-physics responses of the target in various states of burial was also simulated using approaches previously demonstrated as producing a high degree of fidelity to field-collected data.
Finally, comparisons of the response for different unclassified fillers were performed. Consideration of multiple fillers is vital, because it allows the framing of results from the approaches described above so as to avoid making assumptions that are assumed to be universal but are, in fact, specific to a particular filler type.
Even for existing munitions detection systems, which have exhibited promise in the laboratory and in field testing, the relevance of the internal target structures to the received signature (and hence to the choice of detection algorithm feature sets) is largely abstract and based on intuition; systems are treated as black boxes whose robustness can only be estimated empirically. Guidance on the limits of a particular system's robustness is highly valuable and provides researchers with a baseline to improve future algorithm performance. The infrastructure developed in this project and the results obtained identified a number of areas for follow-on work that would provide a deeper understanding of relevant features and how those features contribute to automatic classification performance. Knowledge gained from this SEED effort, specifically the isolation of the elastic response of a target and the various areas of the acoustic template that are indicative of the type of filler will be implemented in SERDP project MR-2505, in which a large focus is the development of physics-based feature sets for automatic target recognition (ATR) algorithms.