Radar Object Classification Using Temporal and Correlation Signatures
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing radar systems struggle to accurately classify objects, especially when they are proximate to each other, due to limitations in the range, angle, or velocity information provided, necessitating improved object classification accuracy.
Innovation Solution
A method combining deep neural networks (DNN) and temporal deep neural networks (TDNN) processes radar samples to generate object signatures by concatenating inter-parameter correlation and temporal information, overcoming overfitting with a compact network design.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional radar systems use only range, angle, or velocity information for object classification, then the system complexity remains low, but the object classification accuracy deteriorates, especially when objects are proximate to each other
Solution Approach 1:
The patent transitions from traditional radar parameters (range, angle, velocity) to a new dimensional space of signal characteristics by applying deep neural networks and temporal deep neural networks. This transforms the classification problem from 3-dimensional parameter space to a high-dimensional feature space, enabling better separation of proximate objects while maintaining manageable system complexity through automated feature extraction.
Solution Approach 2:
The patent introduces neural networks as intermediary components between the radar signal reception and object classification. These neural networks act as mediators that automatically extract and transform raw radar signals into meaningful object signatures, reducing the need for manual feature engineering and lowering overall system complexity while improving classification accuracy.
2Measurement precision
If deep neural networks are used to process radar samples and generate object signatures, then object classification accuracy improves, but power consumption and network size increase
Solution Approach 1:
The patent extracts only the essential temporal and inter-parameter correlation features from radar signals using specialized neural network architectures. By taking out only the critical information needed for classification (temporal behavior patterns and parameter correlations) rather than processing all raw data, the system achieves high accuracy while reducing computational load and power consumption.
Solution Approach 2:
The patent transforms the classification approach by changing from processing raw radar parameters to processing extracted feature representations. The neural networks transform input parameters into compact object signatures that capture essential characteristics, reducing the dimensionality and computational requirements while maintaining or improving classification performance.
3Measurement precision
If complex neural network architectures are used to process temporal radar information, then object classification accuracy improves, but overfitting increases and model generalization deteriorates
Solution Approach 1:
The patent segments the neural network processing into distinct functional components: temporal deep neural networks for extracting temporal behavior patterns, and deep neural networks for capturing inter-parameter correlations. This segmentation allows each component to specialize in specific feature extraction tasks, improving overall accuracy while reducing overfitting through modular, interpretable processing stages.
Solution Approach 2:
The patent employs dynamic processing where temporal deep neural networks adaptively capture time-varying characteristics of objects. The networks dynamically adjust to learn temporal patterns from sequential radar samples, enabling the system to generalize better to unseen objects while maintaining high classification accuracy through adaptive feature learning rather than rigid rule-based approaches.
Data Source
AI summary
A method for radar based object classification, the method may include obtaining multiple radar samples of an object; the multiple radar samples were acquired at different acquisition times; wherein the multiple radar samples comprise a plurality of first radar sample parameters; calculating second radar sample parameters for the multiple radar samples, by applying one or more non-linear functions on at least some of the plurality of first radar sample parameters of at least some of the multiple radar samples; generating an object signature that comprises temporal information and inter-parameter correlation information; wherein the generating comprises feeding, to each one of a deep neural network (DNN) and a time delay neural network (TDNN), (a) at least some of the plurality of first radar sample parameters, and (b) at least some of the second radar sample parameters; and classifying, by a classifier, the signature to a signature class.