Three-dimensional multi-source localization of underwater objects using convolutional neural networks for artificial lateral lines

Abstract

<jats:p>This research focuses on the signal processing required for a sensory system that can simultaneously localize multiple moving underwater objects in a three-dimensional (3D) volume by simulating the hydrodynamic flow caused by these objects. We propose a method for localization in a simulated setting based on an established hydrodynamic theory founded in fish lateral line organ research. Fish neurally concatenate the information of multiple sensors to localize sources. Similarly, we use the sampled fluid velocity via two parallel lateral lines to perform source localization in three dimensions in two steps. Using a convolutional neural network, we first estimate a two-dimensional image of the probability of a present source. Then we determine the position of each source, via an automated iterative 3D-aware algorithm. We study various neural network architectural designs and different ways of presenting the input to the neural network; multi-level amplified inputs and merged convolutional streams are shown to improve the imaging performance. Results show that the combined system can exhibit adequate 3D localization of multiple sources.</jats:p>

The Quadrature Method: A Novel Dipole Localisation Algorithm for Artificial Lateral Lines Compared to State of the Art

“…Yang et al [ 12 ] used a sensor array consisting of two orthogonal lines on a cylinder to demonstrate LCMV beamforming’s 3D localisation performance. Wolf et al [ 15 ] used two parallel ALLs to localise multiple simultaneous sources in 3D. Analysing QM’s performance on 3D localisation tasks would be an interesting future research project.…”

“…Some previous attempts have been made to analytically determine a 2D position and movement direction of a dipole source from its generated fluid flows [ 5 , 25 ]. This task is called the inverse problem of hydrodynamic source localisation [ 15 ]. An intrinsic difficulty of the inverse problem using the dipole field is that both and consist of a directional-dependent combination of two of the three basis wavelets (see Section 2.5.4 ): with where is the radius of the source, is the movement velocity of the source, is the relative position of the sensor from the perspective of the source , and is the azimuth angle of the motion.…”

“…These algorithms attempt to locate objects that move underwater using the water flow pattern their movement generates. This process is analogous to solving the inverse problem of hydrodynamic source localisation [ 15 ]. It is challenging to compare these algorithms’ performance from their original works due to differences in experimental designs and conditions.…”

A Feed-Forward Neural Network Approach for Energy-Based Acoustic Source Localization

“…However, its major drawback lies in their training stage, where the physical presence of training devices is necessary and tedious. Further, Convolutional Neural Network designs were studied with application to localization, by simulating hydrodynamic flow caused by target objects of location, with meritorious accuracy results [19]. Hence, although the application of neural networks can be seen as an option for localization problems, challenges remain to be overcome with regard to their training and structure.…”

Optimal Flow Sensing for Schooling Swimmers

“…Other studies have focused on dipole sources in order to develop methods that extract information and optimize the parameters of the sensing devices [49,50]. In a recent study artificial neural networks were employed to classify the environment using flow-only information [51][52][53][54]. In order to find effective sensor positions weight analysis algorithms were employed [55].…”