SMAUG #tech-Tech Unveiled: Acoustic Detection
SMAUG’s WP4 (Work Package 4) is focused on the underwater detection and location of potential threats such as underwater vehicles, structures, or anomalous devices. This detection and location will be performed using four primary methods:
-
- i) Acoustic Detection
-
- ii) Rapid Sonar Hull Scan
-
- iii) High Resolution Sonar Inspection
-
- iv) Collective Autonomous Location
In this blog article, we will explore the Acoustic Detection technology. What is the methodology behind this technology? And how will this technology be applied and integrated into the SMAUG project? Embark on this journey to explore SMAUG’s tech!
Progress in Acoustic Detection Technologies
The SMAUG project has significantly progressed in developing advanced acoustic detection technologies for maritime applications. Utilising underwater microphones (hydrophones), researchers aim to detect and classify vessels by analysing their sound signatures. Initial tests in controlled environments demonstrated promising results in both indoor and outdoor trials. Real-world trials highlighted challenges with sensitivity over long distances, prompting the development of more refined, sensitive equipment.
To address these challenges, a comprehensive processing chain for the detection and classification system using hydrophones has been proposed. This system covers key workflow components, from acoustic signal reception to classification. A working detection algorithm utilising existing datasets has been developed, but the implementation of a localisation component remains a priority for the next phase. Furthermore, a new deep-learning classification scheme is under development. Early results, particularly from ResNet-based architectures, have shown potential in detecting and classifying various vessels under challenging conditions. Planned improvements include introducing denoising networks or traditional algorithms to further enhance performance. Upcoming experiments with real samples collected using project equipment will focus on refining signal preprocessing, a critical factor for improving detection accuracy.
Testing and Challenges
Initial testing has included indoor trials in controlled environments, such as swimming pools, where robots served as acoustic targets. These tests demonstrated the system’s capability to detect small acoustic sources. However, outdoor tests in Valencia port revealed limitations in hydrophone sensitivity over longer distances. Future efforts will focus on gathering more extensive real-world data using the project’s hardware and publicly available datasets for validation and comparison. To enhance localisation accuracy, a Time Difference of Arrival (TDOA) scheme will be implemented, with various algorithms being explored to estimate target positions using acoustic data. Deep learning classification models will undergo further training with larger and more diverse datasets to boost accuracy, particularly in complex and noisy environments. Additionally, new methods for improving signal processing are under consideration.
Future Plans
Further indoor and outdoor tests are planned to refine hardware configurations and optimise real-world data collection. These trials will include the integration of cameras to provide visual references for vessel activity, as demonstrated in the latest recordings from Valencia port. By addressing these challenges and advancing its methodologies, the SMAUG project is poised to contribute significantly to acoustic detection and classification in maritime applications.
Stay tuned to SMAUG via its social media pages, and don’t forget to subscribe to our newsletter!