Automotive Radar Object Classification with Clutter Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The processing capability of automotive radar sensors for object classification is lacking, particularly in filtering noise and clutter from radar data.
Innovation Solution
A system and method utilizing a convolutional network to extract features, a deconvolutional network for classification, a fully connected network to determine a clutter threshold, and a filter to remove noise and clutter, with optional skip connections for efficient object classification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a convolutional network and deconvolutional network are used for object classification, then classification accuracy is improved, but processing complexity increases
Solution Approach 1:
The system segments the classification task into multiple specialized networks: a convolutional network for feature extraction, a deconvolutional network for classification, and a fully connected network for clutter threshold determination. Each network handles a specific aspect of the classification process, improving overall accuracy while distributing computational complexity across modular components.
Solution Approach 2:
The convolutional network performs preliminary feature extraction from raw radar data before the classification stage. By pre-processing the data to extract relevant features, the system reduces the complexity of the subsequent classification task while maintaining high classification accuracy.
2Reliability
If clutter filtering is applied to radar data, then classification reliability is improved, but processing time increases
Solution Approach 1:
The fully connected network determines clutter threshold values in advance from the radar data before the main classification process. By pre-calculating these thresholds, the filter can quickly remove clutter during classification without adding significant processing time, thereby improving reliability efficiently.
Solution Approach 2:
The filter acts as an intermediary component between the deconvolutional network and the final classification output. It uses clutter threshold values to selectively remove clutter and noise from the classification results, improving reliability while maintaining processing efficiency through targeted filtering rather than comprehensive re-processing.
3Measurement precision
If multiple neural networks are used for feature extraction and classification, then object classification precision is improved, but system complexity increases
Solution Approach 1:
The system divides the classification function into three specialized networks: convolutional network for feature extraction, deconvolutional network for classification, and fully connected network for threshold determination. This segmentation allows each network to be optimized for its specific function, improving overall precision while managing complexity through modular design.
Solution Approach 2:
The bus interconnecting the networks enables a single data structure to serve multiple purposes: it carries feature extraction outputs from the convolutional network, feeds classification inputs to the deconvolutional network, and provides threshold determination inputs to the fully connected network. This multi-functional data pathway reduces system complexity by using shared infrastructure.
Data Source
Figure 1
Figure 2A
Figure 2B
AI summary
A system and method to classify objects using radar data obtained by an automotive radar. The system includes a convolutional network having a plurality of hidden layers comprising convolution layers for extracting features from the radar data, and an output. The system also includes a deconvolutional network having a plurality of hidden layers comprising deconvolution layers for classifying the features extracted from the radar data, and a classification output. The system also includes a filter having an input coupled to the classification output of the deconvolutional network. The system further includes a fully connected network having a plurality of fully connected layers for determining a clutter threshold value from the output of the convolutional network. The filter is operable to use the clutter threshold value to filter noise and/or clutter from the classification output of the deconvolutional network and pass a filtered classification output to an output of the system.