A Deep-Learning Model for Underwater Position Sensing of a Wake's Source Using Artificial Seal Whiskers

Area of Science:

• Biomimetics and Sensor Technology
• Fluid Dynamics
• Machine Learning

Background:

• Marine animals utilize hydrodynamic sensing for navigation and prey detection in aquatic environments.
• Artificial systems can mimic these biological capabilities for underwater sensing applications.

Purpose of the Study:

• To develop and evaluate a biomimetic sensor for underwater position detection of a wake-generating body.
• To apply deep learning techniques to relate sensor vibrations to object location.

Main Methods:

• A sensor composed of optical fibers mimicking seal whiskers was used.
• Supervised learning related artificial whisker deflections to an upstream cylinder's position.
• Two neural network models were trained and validated using 10-fold cross-validation.

Main Results:

• The deep-learning models accurately predicted the 2D coordinates of the cylinder.
• The sensor achieved high accuracy in estimating the cylinder's position at Re ≃ 6000.
• Position estimation error was less than the cylinder's diameter for distances up to 25 times the diameter.

Conclusions:

• The biomimetic sensor effectively detects and localizes underwater objects using hydrodynamic sensing.
• Deep learning provides a powerful tool for interpreting complex sensor data in fluid environments.
• This technology has potential applications in underwater navigation and surveillance.