3D Forward-Looking Sonar Recognition Using CNN-Based Target Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing three-dimensional forward-looking sonar systems generate large amounts of data that are difficult for human operators to interpret in real-time, limiting their application for safe navigation and target detection in water.
Innovation Solution
Implementing machine learning algorithms, specifically convolutional neural networks, to process and classify three-dimensional sonar data, incorporating additional data sources for training to improve accuracy and reduce false positives and negatives.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If three-dimensional forward-looking sonar systems generate large amounts of data for improved detection accuracy, then target detection precision is improved, but data interpretation difficulty increases
Solution Approach 1:
The patent introduces machine learning algorithms as an intermediary between the sonar system and the operator. The ML model processes the complex 3D sonar data and generates simplified, actionable outputs such as target classifications and seafloor mapping, making the data interpretable for navigation decisions without requiring operators to manually analyze the raw volumetric data.
Solution Approach 2:
The patent replaces manual mechanical data interpretation with automated machine learning processing. Traditional manual analysis of 3D sonar data is substituted by neural networks that automatically detect, classify, and map underwater features, eliminating the need for operators to manually process complex volumetric datasets.
2Ease of operation
If traditional detection algorithms are used for simplicity, then ease of operation is maintained, but detection accuracy and reliability deteriorate
Solution Approach 1:
The patent changes the fundamental parameters of data processing by transitioning from traditional signal processing methods to machine learning-based approaches. The system uses deep learning models that can process complex 3D volumetric data and provide more reliable detection accuracy while maintaining ease of operation through automated processing.
Solution Approach 2:
The patent creates a digital copy of the sonar data processing function through machine learning models. Instead of operators manually analyzing data, the system uses trained neural network models that replicate and enhance human detection capabilities, providing consistent and reliable results without operator fatigue or subjectivity.
3Measurement precision
If machine learning models with more layers and parameters are used to improve detection accuracy, then measurement precision is improved, but computational complexity increases
Solution Approach 1:
The patent segments the machine learning model into specialized components, including separate networks for seafloor detection, in-water target detection, and classification. This segmentation allows each component to be optimized for its specific function while maintaining overall system manageability and reducing unnecessary computational complexity.
Solution Approach 2:
The patent applies three-dimensional convolutional neural networks that process volumetric sonar data in 3D space. This dimensional approach allows the model to capture spatial relationships and depth information more effectively than traditional 2D methods, improving measurement precision for seafloor depth determination while working with the inherent 3D structure of the data.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enables real-time, precise, and accurate detection and classification of seafloor and in-water targets, enhancing navigation safety and usability of sonar systems.
Implementation Method 1
Sonar systems use sound propagation to map the seafloor and identify objects under the surface of the water
Implementation Method 2
sonar sensor's data... three-dimensional array of backscatter strength from the sonar sensor
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Machine learning algorithms can interpret three-dimensional sonar data to provide more precise and accurate determination of seafloor depths and in-water target detection and classification. The models apply architectures for interpreting volumetric data to three-dimensional forward-looking sonar data. A baseline set of training data is generated using traditional image and signal processing techniques, and used to train and evaluate a machine learning model, which is further improved by additional inputs to improve both seafloor and in-water target detection.