Radar Object Recognition Using Reflection Distribution Modeling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional radar image processing and histogram analysis technologies require large amounts of data for accurate object identification, are sensitive to signal measurement environments, and struggle to distinguish between similar reflective surface materials, leading to limitations in performance and hardware costs for autonomous driving systems.
Innovation Solution
A method and apparatus that model radar reflection signal data using a mixed normal distribution, allowing for efficient recognition of moving objects by analyzing the distribution characteristics of received signal intensities, which reduces data requirements and hardware costs while improving accuracy and extensibility for vehicle collision prevention systems.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If deep learning technology is used for radar image processing, then object identification accuracy is improved, but a large amount of learning data is required
Solution Approach 1:
The patent segments the object identification process into two distinct stages: first clustering radar reflection signal data by material type using unsupervised learning, then classifying objects by type using supervised learning with significantly less data. This segmentation allows each stage to use the most appropriate learning method, reducing overall data requirements while maintaining accuracy.
Solution Approach 2:
The patent introduces an intermediate step of material-based clustering between raw radar data and final object classification. This intermediary process groups similar reflection characteristics together, creating more efficient training data for the subsequent classification stage and reducing the amount of labeled data needed.
2Extent of automation
If end-to-end deep learning method is used, then object identification is automated, but it becomes difficult to analyze the cause of identification performance
Solution Approach 1:
The patent divides the automated identification process into separable stages with distinct functions: material clustering and type classification. This segmentation enables independent analysis of each stage's contribution to overall performance, making it easier to identify causes of identification issues while maintaining automation.
Solution Approach 2:
The patent implements feedback mechanisms where clustering results inform classification processes, and performance analysis feeds back into system optimization. This structured feedback enables systematic analysis of identification causes while preserving automated operation.
3Measurement precision
If radar image histogram analysis technology is used, then reflection characteristic modeling is achieved, but a lot of data is required for modeling
Solution Approach 1:
The patent applies partial action by using unsupervised learning for material clustering without requiring extensive labeled data, then applying supervised learning only for the specific classification task. This partial application of different learning methods reduces overall data requirements compared to comprehensive supervised learning.
Solution Approach 2:
The patent changes the learning approach parameter from supervised to unsupervised for the modeling stage, allowing the system to learn reflection characteristic patterns without requiring large amounts of labeled training data, thereby achieving accurate modeling with less data.
4Device complexity
If morphology operation is applied for clustering, then image processing is simplified, but small-sized moving objects may not be identified
Solution Approach 1:
The patent changes the fundamental approach parameter from morphology-based processing to radar reflection signal-based clustering. This parameter change enables the system to identify small objects through their unique reflection characteristics rather than relying on size-based morphological operations, maintaining processing simplicity while improving detection accuracy.
Data Source
AI summary
A method for recognizing an object comprises storing a reference reflection characteristic for each of classes based on modeling radar reflection signal data for each of objects corresponding to each of the classes, determining, among the classes, a class of a reference reflection characteristic of a high similarity with a reflection characteristic of received signal data transmitted from a radar, and identifying a target object of the received signal data based on the determined class and outputting information of the target object.